1

In situ multiple sulfur isotope analysis by SIMS of pyrite, chalcopyrite, 1

pyrrhotite, and pentlandite to refine magmatic ore genetic models 2

3

LaFlamme, Crystala*, Martin, Laureb, Jeon, Heejinb, Reddy, Steven, M.c, Selvaraja, 4

Vikramana, Caruso, Stefanoa, Bui, Thi Haod, Roberts, Malcolm P.b, Voute, Francoisa, 5

Hagemann, Steffena, Wacey, Davidb, Littman, Stene, Wing, Boswelld, Fiorentini, Marcoa, 6

Kilburn, Matthew R.b 7

aCentre for Exploration Targeting, ARC Research Council Centre of Excellence for Core to 8

Crust Fluid Systems (CCFS), University of Western Australia, Australia 9

bCentre for Microscopy, Characterisation, and Analysis, ARC Centre of Excellence for Core 10

to Crust Fluid Systems (CCFS), University of Western Australia, Australia 11

cThe Institute for Geoscience Research, Department of Applied Geology, Curtin University 12

of Technology, Australia 13

dDepartment of Earth and Planetary Sciences and GEOTOP, McGill University, Canada 14

eMax-Planck-Institut für Marine Mikrobiologie, Bremen, Germany 15

16

17

*corresponding author: crystal.laflamme@uwa.edu.au 18

19

ABSTRACT 20

With growing interest in the application of in situ multiple sulfur isotope analysis to a variety 21

of mineral systems, we report here the development of a suite of sulfur isotope standards for 22

distribution relevant to magmatic, magmatic-hydrothermal, and hydrothermal ore systems. 23

These materials include Sierra pyrite (FeS2), Nifty-b chalcopyrite (CuFeS2), Alexo pyrrhotite 24

(Fe(1-x)S), and VMSO pentlandite ((Fe,Ni)9S8) that have been chemically characterized by 25

electron microprobe analysis, isotopically characterized for δ33S, δ34S, and δ36S by 26

fluorination gas-source mass spectrometry, and tested for homogeneity at the micro-scale by 27

secondary ion mass spectrometry. Beam-sample interaction as a function of crystallographic 28

orientation is determined to have no effect on δ34S and Δ33S isotopic measurements of 29

pentlandite. These new findings provided the basis for a case study on the genesis of the 30

Long-Victor nickel-sulfide deposit located in the world class Kambalda nickel camp in the 31

southern Kalgoorlie Terrane of Western Australia. Results demonstrate that precise multiple 32

sulfur isotope analyses from magmatic pentlandite, pyrrhotite and chalcopyrite can better 33

constrain genetic models related to ore-forming processes. Data indicate that pentlandite, 34

pyrrhotite and chalcopyrite are in isotopic equilibrium and display similar Δ33S values 35

+0.2‰. This isotopic equilibrium unequivocally fingerprints the isotopic signature of the 36

2

magmatic assemblage. The three sulfide phases show slightly variable δ34S values 1

(δ34Schalcopyrite = 2.9 ± 0.3‰, δ34Spentlandite = 3.1 ± 0.2‰, and δ34Spyrrhotite = 3.9 ± 0.5‰), which 2

are indicative of natural fractionation. Careful in situ multiple sulfur isotope analysis of 3

multiple sulfide phases is able to capture the subtle isotopic variability of the magmatic 4

sulfide assemblage, which may help resolve the nature of the ore-forming process. Hence, 5

this SIMS-based approach discriminates the magmatic sulfur isotope signature from that 6

recorded in metamorphic- and alteration-related sulfides, which is not resolved during bulk 7

rock fluorination analysis. The results indicate that, unlike the giant dunite-hosted komatiite 8

systems that thermo-mechanically assimilated volcanogenic massive sulfides proximal to 9

vents and display negative Δ33S values, the Kambalda ores formed in relatively distal 10

environments assimilating abyssal sulfidic shales. 11

12

HIGHLIGHTS 13

Characterisation of four sulfide standards for multiple sulfur isotope analysis: pyrite, 14

chalcopyrite, pyrrhotite, and pentlandite for distribution 15

Analysis of orientation effect in pentlandite 16

Natural sulfur isotope fractionation between pentlandite and pyrrhotite 17

Case study multiple sulfur isotope analysis of three sulfide phases within world-class 18

Long-Victor komatiite-hosted nickel-sulfide deposit 19

20

KEYWORDS 21

Multiple sulfur isotopes, SIMS, in situ, sulfide minerals, ore genesis 22

23

1. INTRODUCTION 24

Sulfur is a trace element in silicate melts, typically concentrated below 0.2 wt.%. However, it 25

is an essential element in a wide range of environments including the lithosphere, biosphere, 26

hydrosphere, and atmosphere. In recent years, our understanding of the sulfur cycle and its 27

role in the evolution of these terrestrial reservoirs has been revolutionised by the study of the 28

sulfur isotope composition of pyrite, the most common sulfide mineral (Farquhar et al., 2000; 29

Kump, 2012 and references therein; Strauss, 1997; Thomassot et al., 2015). We have gained a 30

fundamental understanding into the development of early Earth’s processes, in particular 31

those linked to the emergence of life and the development of an oxygenated atmosphere 32

(Farquhar et al., 2000), by the discovery of mass independent fractionation (MIF) of sulfur 33

isotopes. 34

35

Sulfur resides in the Earth’s mantle, crust and hydrosphere but is locally concentrated in 36

mineralised systems typically associated with ore deposits, where it acts as the primary 37

complexing ligand in the formation of sulfide minerals. Mantle- and crustally-derived 38

magmas have brought large quantities of economic metals from the Earth’s interior to the 39

3

near surface, and hydrothermal fluids have remobilised and re-precipitated these metals 1

within the crust as different sulfides. The sulfur itself may be sourced from a variety of 2

compositional reservoirs, each with distinct isotopic compositions. Mixing and interactions 3

with the mantle, crustal magmas, hydrothermal fluids, country rocks, or meteoric waters 4

imparts specific isotopic signatures, resulting in minerals with a range of isotopic 5

compositions. As such, intra-grain and inter-grain chemical and isotopic variations in sulfur-6

rich mineralised systems record the interaction of these different reservoirs and offer unique 7

insights into the complex fluid-rock interactions within mineral systems (McCuaig et al. 8

2010). For example, in magmatic ore deposits, sulfur isotope data have fingerprinted the 9

source of the sulfur linked to ore genesis (Bekker et al., 2009; Chang et al., 2008; Fiorentini 10

et al., 2012a; Hiebert et al., 2013; Lesher and Groves, 1986; Penniston-Dorland et al., 2008; 11

Sharman et al., 2013) and constrained the geodynamic framework where these deposits 12

formed (e.g., Chen et al., 2015; Fiorentini et al., 2012b; Giacometti et al., 2014). Similarly, 13

sulfur isotope studies have proven to be vital in characterising magmatic-hydrothermal (Helt 14

et al., 2014; Xue et al., 2013) and hydrothermal systems (e.g., Jamieson et al., 2013; Leach et 15

al., 2005; Sharman et al., 2015). Constraining the sulfur isotopic signature in magmatic-16

hydrothermal mineral systems is useful in delineating the source of sulfur, and is an 17

important parameter to understand how, when and where sulfur saturation occurs (e.g., Evans 18

et al., 2014). In addition, such data provides a better understanding of the geodynamic 19

environment in which the mineralising process occurs which impacts on the targeting 20

rationale applied during exploration (e.g., Fiorentini et al., 2012a). Consequently, ore 21

deposits are a perfect laboratory for understanding the source and mobility of sulfur in a wide 22

variety of settings. 23

24

Mineral systems and ore deposits have characteristically complex microscale intra-granular 25

and inter-granular textures due to variations in their chemistry during formation and 26

subsequent re-equilibration during cooling (e.g., pentlandite exsolution in pyrrhotite; Durazzo 27

and Taylor, 1982). In situ sulfur isotope analysis at the microscale has the potential to 28

revolutionise our understanding of ore forming processes. The development of in situ 29

analytical techniques using laser ablation-(multi-collection)-inductively coupled plasma mass 30

spectrometry (LA-MC-ICPMS; see Bühn et al., 2012; Craddock et al., 2008) and large 31

geometry secondary ion mass spectrometry (SIMS; see Farquhar et al., 2013; Ireland et al., 32

2014; Ushikubo et al., 2014; Whitehouse, 2013) now allows high-precision isotopic analysis 33

of multiple sulfur isotopes with spot sizes nearing ten microns. The ultra-high sensitivity of 34

SIMS in particular affords the ability to measure the least common stable isotopes of sulfur, 35 33S (0.75%) and 36S (0.02%), together with the more abundant 32S (95.02%) and 34S (4.21%) 36

isotopes, simultaneously from the same volume of material. This gives the potential to 37

identify the anomalous sulfur isotopic signatures indicative of mass independent fractionation 38

(MIF; Δ33S and Δ36S) together with δ34S. 39

40

Instrumental mass fractionation in SIMS is intrinsically linked to the composition and 41

crystallographic orientation of the material being analysed and the specific conditions under 42

which the analysis is performed. Therefore, accurate isotopic measurements require careful 43

standardization against a suitable matrix-matched reference material (Eiler et al., 1997; Stern, 44

4

2008). Although a number of reference materials have been developed for acquiring in situ 2-1

sulfur isotopes (e.g., Kozdon et al., 2010), the majority of published in situ 3- or 4-sulfur 2

isotope analyses have been acquired from the most common sulfide mineral, pyrite. As such, 3

many SIMS laboratories worldwide have developed “in-house” pyrite (and to a lesser degree 4

other) standards for multiple sulfur isotope analysis (Balmat, Isua 248474, Ruttan; see 5

Whitehouse, 2013; Ushikubo et al., 2014; Hauri et al., 2016). With growing interest in the 6

application of in situ sulfur isotope analysis to a wide variety range of mineral systems, there 7

is an increased need for reference material for a variety of common sulfide minerals, which 8

presently remain scarce. In addition to composition, sulfide crystallography has previously 9

been demonstrated to potentially create an orientation effect-induced instrumental mass bias 10

on sulfur isotope analysis for certain sulfides including galena (PbS) and sphalerite 11

((Zn,Fe)S) but not others including pyrite (FeS2), pyrrhotite (Fe(1-x)S), and chalcopyrite 12

(CuFeS2) (Kozdon et al., 2010; Kita et al., 2011). Of these sulfides, most have a diamond-13

cubic crystal structure, with the exception of galena which has a cubic hexoctahedral 14

structure. Here, we build on this observation to demonstrate that orientation effect does not 15

induce instrumental mass fractionation on δ34S and Δ33S in another cubic hexoctahedral 16

sulfide - pentlandite ((Fe,Ni)9S8). 17

18

Deviations from mass dependent fractionation are typically (but not uniquely) triggered by 19

the presence of an oxygen-poor atmosphere that existed before the Great Oxidation Event at 20

ca. 2.4 Ga in which ultraviolet radiation was the driver for mass independent photochemical 21

separation of sulfur isotopes (e.g., Farquhar et al., 2000; Farquhar and Wing, 2003). For this 22

reason, the measure of 33S has become essential to evaluate the full suite of isotopic features 23

of Archean rocks (e.g., Bühn et al., 2012; Farquhar et al., 2013), and those from younger 24

terranes that might be sourcing Archean rocks (e.g., Cabral et al., 2013; Selvaraja et al., in 25

submission). The least abundant stable isotope of sulfur, 36S, behaves similarly to 33S, and has 26

also become increasingly important for fingerprinting Archean source rocks and 27

understanding early Earth processes. The Δ36S/Δ33S ratio can elucidate between mass 28

dependent processes and contributions from Archean MIF sources when magnitudes of mass 29

dependent fractionation deviations are small (Farquhar et al., 2007; Johnston, 2011). 30

31

We present multiple sulfur isotope data from pentlandite, pyrrhotite, and chalcopyrite from 32

an Archean komatiitic massive nickel-sulfide deposit. To do so, we have developed four 33

reference materials (pyrite, chalcopyrite, pyrrhotite, and pentlandite) that we have fully 34

chemically and isotopically characterized for multiple sulfur isotope analysis. We present 35

multiple sulfur isotope results from multiple sulfides within the deposit to constrain the 36

source of the sulfur that triggered sulfide saturation and investigate the geodynamic setting in 37

which this magmatic system was emplaced. 38

39

2. METHODS 40

To determine the chemical and isotope composition of potential sulfide reference materials 41

we combined three analytical techniques. Firstly, we performed wavelength dispersive 42

spectrometry (WDS) by electron probe micro-analysis (EPMA) on a multitude of grains (or 43

grain fragments) of each candidate reference material to ensure that the reference material is 44

5

chemically homogeneous across a number of grains (or grain fragments). Chemical 1

compositions (spot analyses) were linked to chemical WDS maps and scanning electron 2

microscope backscatter electron (SEM-BSE) images to highlight potential chemical zonation, 3

mineralogical inclusions, and fractures. Details of SEM-BSE and EPMA-WDS are presented 4

in the Supplementary Material. 5

6

Secondly, at least five grains (or grain fragments) of each candidate reference material were 7

analysed for in situ multiple sulfur isotopes by SIMS to determine whether an adequate level 8

of reproducibility exists at the intra- and inter-grain scale (isotopic homogeneity). Multiple 9

materials were tested prior to selecting the material that displayed adequate isotopic 10

homogeneity (reproducibility on 34S/32S better than 0.5‰). For instance, five materials from 11

different environments were tested to find a suitable chalcopyrite standard (see section 5.1 for 12

discussion on selecting sulfide reference material). To further determine on what scale 13

sufficient homogeneity existed once the reference material was selected, more than 40 grains 14

for each material were analysed. If the reference material was deemed suitable, multiple 15

grains or grain fragments (in the case of material derived from large crystals) were analysed 16

by bulk multiple sulfur analyses. Bulk analyses were completed by fluorination gas-source 17

mass spectrometry. In each case, sulfur was extracted from 5 to 11 separate aliquots (of 18

extracted sulfur from separate grains or grain fragments) and analysed for its multiple sulfur 19

isotopic composition. We further evaluated the homogeneity of the SIMS material by 20

completing an MSWD test on the data after Wing and Farquhar, (2015). The respective 21

analytical techniques are described in detail below. 22

23

For the case study, we investigated the chemical composition and multiple sulfur isotope 24

signature of three phases (pentlandite, pyrrhotite, chalcopyrite). A micro-X-ray Fluorescence 25

(XRF) elemental map was acquired to visually investigate the textural relationship between 26

the three sulfide phases and choose samples for in situ and bulk isotope analyses (details 27

regarding analytical set up are presented in the Supplementary Materials). EPMA analyses on 28

pyrrhotite and pentlandite were completed to ensure a similar chemical composition to the 29

reference materials (see Appendix A). In situ isotope signatures of pentlandite, pyrrhotite and 30

chalcopyrite were compared to ten bulk rock fluorination values that incorporated all the co-31

existing sulfide phases (see Appendix B and C). We also used the case study material to 32

assess the crystallographic orientation effect in pentlandite. To do so, Electron Backscatter 33

Diffraction Analysis (EBSD) was completed on this sample and pentlandite crystal 34

orientation was compared with multiple sulfur isotope results. 35

36

2.1 Secondary Ion Mass Spectrometry 37

2.1.1 Analytical set up and conditions 38

In situ sulfur isotopic ratios were measured using a CAMECA IMS1280 large-geometry ion 39

microprobe at the CMCA, UWA. Sample mounts were made by coring 3 mm diameter pucks 40

from rock fragments using a drill press fitted with diamond drill bits, then mounted and cast 41

in the central portion of a 25 mm in diameter epoxy mount. Standard blocks (cast separately 42

to be reused) were made by mounting 1-2 grain fragments of pyrite chalcopyrite and 1 mm 43

diameter pucks of rock fragments host to pyrrhotite and pentlandite. Reference materials 44

6

were cast in epoxy ~8 mm from the edge of a mount. Sample mounts and standard blocks 1

were trimmed to a thickness of 5 mm using a precision saw, coated with 30 nm of gold, and 2

mounted together (after being appropriately trimmed) in the sample holder (further details 3

provided in the Supplementary Materials). Care was taken to set the surfaces of the standard 4

and sample blocks at the same level in the sample holder. If necessary, a small amount of 5

carbon paint was applied to provide good conductivity between both pieces. To ensure that 6

results are not a product of analytical artefacts due to X-Y-Z positioning, a reference material 7

is also mounted and cast with the sample block to compare with values generated from the 8

standard block. Additional details regarding sample preparation for SIMS analysis are 9

presented in the Supplementary Materials. 10

11

The ion microprobe was operated in multicollection mode using a Cs+ primary beam with an 12

intensity of ~1–4 nA in Gaussian mode that interacted with the sample at 20 keV. In some 13

instances, depending on the nature of the sample (i.e., size of sulfides, surrounding material, 14

whether sulfides are mounted as individual grains or incorporated in rock chips) a normal 15

incidence electron flood gun was used for charge compensation. Following a 30 s pre-sputter, 16

secondary sulfur ions from the target sample were extracted at -10 kV and admitted to the 17

mass spectrometer with a field magnification of 133×, with automated centering of the 18

secondary beam in the field/aperture (both x and y; aperture size 4000 μm) and entrance slit 19

(x direction only; slit width 60 μm or 90 µm in some cases). The NMR magnetic field 20

controller locked the axial mass at the beginning of each session, and the mass spectrometer 21

operated at a mass resolution (M/ΔM) of about 2500 (exit slit width of 500 μm on the 22

multicollector). Under these conditions, the hydrite interference 32S1H on the 33S peak was 23

avoided by offsetting the 33S peak centre to the low mass side. 24

25

For triple sulfur isotope measurements (32S, 33S and 34S), a 15 μm raster was applied, and the 26

sulfur isotopes were simultaneously detected by three Faraday Cups using amplifiers with 27

1010 Ω (L’2), 1011 Ω (L1), and 1011 Ω (FC2 or H1) resistors. Data were collected over 123 s 28

of acquisition time in 20 integration cycles. Count rates on 32S varied from 1×109 to 3.5×109 29

cps on pyrite, depending on the intensity of the primary beam but is typically 2.2–2.4×109 cps 30

with a 2.5 nA primary beam. 31

32

Quadruple sulfur isotope (32S, 33S, 34S, and 36S) analysis has different analytical protocols to 33

include the measurement of 36S using a low-noise ion counting electron multiplier (EM) in 34

the H2 position of the multicollector axis. A higher primary beam current (3–4 nA), larger 35

raster (30 μm for pre-sputter and 20 μm for analysis) and longer acquisition time (279 s in 45 36

integration cycles) were necessary to achieve adequate repeatability on 36S. The count rates 37

on 36S of typically 2–4×105 cps for pyrite cause a significant gain drift for the EM. Hence, the 38

EM high voltage was optimised using a Pulse Height Amplitude (PHA) distribution curve at 39

the start of each session, and the gain drift was measured three times during each analysis (at 40

the beginning, middle and end) and corrected by the CAMECA CIPS software (see 41

Schuhmacher et al., 2004). In the case of unknown material, measurements were interspersed 42

with matrix-matched reference material to calibrate isotope ratios and monitor internal 43

sample repeatability. 44

7

1

2.1.2 Data processing and error propagation 2

In situ sulfur isotope measurements by SIMS are corrected in two steps. Firstly, the pyrite 3

standard (Sierra; see section 3.1) is analysed once every five to eight analyses, regardless of 4

whether the pyrite is being used as the primary standard for that particular session, allowing 5

for the assessment of the stability of the instrument during the given analytical session, and 6

correct for instrumental drift. Secondly, sample isotopic ratios are corrected for instrumental 7

mass fractionation using the correction factor α, determined by normalising the mean of all 8

measurements on the matrix matched reference material, Rstd, to the isotopic ratio of the 9

reference material RRMas obtained by independent bulk methods (e.g. fluorination gas-source 10

mass spectrometry): 11

(1) α =Rstd

RRM 12

The propagated uncertainty for the δxS value of each sample spot takes into account the 13

internal error on the raw isotopic ratios, the uncertainty on the drift correction where 14

necessary and the uncertainty on the standard measurement, calculated as the standard 15

deviation on the mean isotopic ratios measured in the standards. 16

17

Identifying MIF has become increasingly important in assessing a number of geological 18

environments because it reveals fundamental information pertaining to the age and source of 19

sulfide mineralization (e.g., Johnston, 2011). MIF is presented as Δ33S and Δ36S to quantify 20

the deviation from the mass dependant fractionation slope; however, these values can be 21

small in magnitude – often much less than 1.0‰ (see Figure 1). Therefore, it is important to 22

systematically quantify uncertainty on these values to ascertain whether a MIF signature does 23

indeed exist. 24

25

The mass independent relationship is denoted by the Δ33S and Δ36S notation to represent the 26

deviation between the isotopic ratios measured and those predicted according to mass 27

dependent fractionation, and is defined as: 28

(2) ∆xS𝑖 = δxS𝑖 − 1000 × [(δ34SV−CDT

1000+ 1)

λ

− 1] 29

where x is either 33 or 36 and λ is the slope of the mass dependent fractionation line (0.515 30

for δ33S and 1.91 for δ36S; Hulston and Thode, 1965; Ono et al., 2006a). The λ values 31

approximate the relationships for high temperature equilibrium isotopic fractionations 32

(Farquhar and Wing, 2003). 33

8

1

Figure 1: Calculated magnitudes of mass independent fractionation as deviations from the mass dependent 2 fractionation line (MDF) presented in δ33S vs. δ34S space. The small deviations from the MDF line represent 3 mass independent fractionation. This highlights the importance of: 1) precise and accurate multiple sulfur 4 isotope measurements, and 2) quantification of uncertainty on Δ33S. Although not expressed in this figure, 5 quantifying Δ36S is equally important. 6

7

The uncertainty on Δ33S and Δ36S is calculated by propagating the uncertainties on δ33S, δ34S, 8

and δ36S; however, paired sulfur isotope ratios covary in δ33S–δ34S and δ36S–δ34S space. To 9

account for this relationship, we determined the covariance on the reference material of the 10

analytical session and propagated to the measurements of δ33S and δ34S (𝜎δ34Sδ33S) and δ36S 11

and δ34S (𝜎δ34Sδ36S). The uncertainty on ΔxS is calculated here in a similar manner to 12

Farquhar et al. (2013) but also accounts for this covariance in the formulation of the 13

uncertainties on Δ33S and Δ36S as follows: 14

15

(3) 𝜎ΔxS = √𝜎δxS

2 + 𝜎δ34S2 × [λ × (1 +

δ34S

1000)

−(1−λ)

× 𝜎δ34S2 ]

2

+ 2 × 𝜎δ34SδxS ×

[λ × (1 +δ34S

1000)

−(1−λ)

]

16

17

18

where variables are as in equation 2. 19

20

2.2 Fluorination gas-source mass spectrometry 21

Samples for fluorination coupled with gas-source mass spectrometry were analysed at the 22

Stable Isotope Laboratory of the Department of Earth and Planetary Sciences at McGill 23

University, Montreal, Canada. Sulfide-bearing samples were microdrilled using a 1 mm 24

9

diameter hand drill. Each sample was microdrilled multiple times. Sulfur was chemically 1

extracted from the 15–30 mg powders to form silver sulfide by chromium reduction as 2

described by Canfield et al. (1986). Silver sulfide was fluorinated at 225°C in a Ni bomb 3

under F2 atmosphere for nine hours to produce SF6. The samples were purified cryogenically 4

and by gas chromatography and introduced by SF6 line into a Thermo Electron MAT 253 5

mass spectrometer fitted with a dual inlet to measure 32SF5+, 33SF5

+, 34SF5+, and 36SF5

+. Sulfur 6

isotopic ratios are expressed on the V-CDT scale, on which the δ34S, Δ33S, and Δ36S values of 7

the Ag2S reference material, IAEA-S-1, are taken to be -0.3‰, 0.094‰, and -0.7‰, 8

respectively (Wing and Farquhar, 2015). The precision and accuracy of the bulk fluorination 9

system is evaluated by repeat analyses that return uncertainty (2SD) on δ34S, Δ33S, and Δ36S 10

values as better than 0.15‰, 0.02‰ and 0.4‰, respectively. The Δ33S and Δ36S values are 11

calculated in the same manner as described in equation 2. 12

13

2.3 Electron Backscatter Diffraction Analysis (EBSD) 14

EBSD analysis of pentlandite to compare crystallographic orientation with multiple sulfur 15

isotope analysis was undertaken on a Tescan MIRA3 field emission SEM at the Microscopy 16

and Microanalysis Facility, John de Laeter Centre, Curtin University, Australia. Pentlandite 17

from massive sulfide ore was analysed within four pucks that were prepared into a mount as 18

described in the Supplementary Materials. Following standard petrographic diamond 19

polishing, the mount was polished chemically and mechanically to 60 nm using colloidal 20

silica in pH10 NaOH. Prior to analysis the sample was coated with a 5 nm carbon film. 21

EBSD data were acquired from a sample titled at 70° using an Oxford Instruments Aztec 3.0 22

system operating at 12 kV, a beam intensity of 16 nA and fixed working distance of 20 mm. 23

Pentlandite, pyrrhotite and magnetite orientation data were collected from each sample. 24

Crystallographic data required to create the theoretical match units, by which empirically 25

collected patterns are compared, were derived from Rajamani and Prewitt (1975) for 26

pentlandite, Wechsler et al. (1984) for magnetite and Alsen (1925) for pyrrhotite. EBSD data 27

were collected using the “mapping” mode of the Aztec EBSD software with a step size of 10 28

µm, a minimum of 8 bands and a Hough resolution of 60. In all cases, solutions gave mean 29

angular deviation (MAD) values of <1° and calculation intergrain orientations are 30

reproducible. 31

32

Post-processing of EBSD data was undertaken using Oxford Instruments Channel 5.12 33

software. Noise reduction protocols applied to the data were the Channel “wildspike” 34

correction and a 5 nearest neighbour zero solution algorithm. EBSD data are shown as 35

standard phase and inverse pole figure maps in the Supplementary Materials. 36

37

3. MATERIALS 38

39

Sulfides are inherently heterogeneous both chemically and mineralogically. Therefore, 40

identifying natural specimens that may be used as sulfur isotope reference material is a 41

difficult task whereby geological environment, pressure-temperature conditions, chemical 42

composition, and grain size need to be considered. In the case of selecting reference material 43

for pyrite and chalcopyrite, small pieces (~2 cm3) of much larger crystals from hydrothermal 44

10

deposits demonstrate a more reasonable degree of chemical and isotopic homogeneity over 1

small individual crystals that make up massive to sulfide-rich layers. For each of these 2

sulfides, although the total volume amount of reference material characterized is small (in 3

order to ascertain isotopic homogeneity), advancements in SIMS and LA-ICPMS capabilities 4

and procedures (i.e., new sample holder geometry in SIMS; Peres et al. 2012; and large 5

volume cells in LA-ICPMS) allow for the repeat use of the same reference material mounted 6

in standard blocks. 7

8

In the case of pyrrhotite and pentlandite, these two sulfide phases are commonly exsolved 9

from a monosulfide solid solution and typically have intra-grain associations in magmatic 10

environments (Durazzo and Taylor, 1982; Kelly and Vaughn, 1983). Furthermore, the 11

chemical compositions of pyrrhotite (Fe(1-x)S; x = 0.0–0.2) and pentlandite (Fe, Ni)9S8 are 12

variable, reflective of conditions under which they formed. Pyrrhotite and pentlandite do not 13

commonly form large hydrothermally-derived crystals, and so characterized reference 14

material is derived from a small piece of a sulfide-rich and massive sulfide horizon, 15

respectively. Images of the reference materials are presented in the Supplementary Materials. 16

The effects of intra-grain and inter-grain textures on isotopic and chemical homogeneity of 17

pyrrhotite and pentlandite reference materials are assessed in sections 4.3 and 4.4. Additional 18

discussion on selecting reference materials is presented in section 5.1. 19

20

3.1 Pyrite – Sierra 21

Sierra is a 2 cm3 cube from a large 9 kg cube of pyrite sourced from a mine in the Cretaceous 22

stratigraphy of the Sonora region of Mexico. Although the exact provenance is unknown, the 23

area experienced widespread igneous activity during the late Cretaceous, resulting in the 24

formation of widespread porphyry copper mineralisation across the region (Barra and 25

Valencia, 2014). Sierra is mounted as 0.5 mm wide fragments of pyrite from the 2 cm3 cube 26

of Sierra for SIMS analysis. 27

28

3.2 Chalcopyrite – Nifty-b 29

The Nifty-b is a 2 cm3 piece of a larger 8 cm3 grain of chalcopyrite from the Nifty copper 30

deposit in the Proterozoic Paterson Orogen of Western Australia. Mineralization is hosted by 31

the Nifty stratigraphic member of the Yeneena Supergroup and occurs as hydrothermal 32

chalcopyrite-quartz-dolomite replacement of low grade shale (Anderson et al., 2001). Nifty-b 33

is mounted as 0.5 mm wide fragments of chalcopyrite pyrite from the 2 cm3 piece of Nifty-b 34

for SIMS analysis. 35

36

3.3 Pyrrhotite – Alexo 37

The Alexo monoclinic pyrrhotite forms subhedral 0.1–2 mm grains within a matrix of 38

peridotite. It is from the nickel-sulfide Alexo deposit within the ca. 2.7 Ga Abitibi granite-39

greenstone belt of the Superior Craton, Canada. The sample comprises magmatic pyrrhotite 40

grains composed of disseminated to net-textured sulfides at the contact between olivine 41

cumulate komatiite rocks of the Munro Group and footwall andesitic rocks of the Hunter 42

Mine Group (Naldrett, 1966). The local metamorphic grade at Alexo is prehnite-pumpellyite-43

facies. Based on stratigraphic associations, the timing of the nickel-sulfide mineralization at 44

11

Alexo is assumed to have occurred at ca. 2.7 Ga (Fyon and Green, 1991), which is coeval 1

with other world-class komatiite-hosted nickel-sulfide systems worldwide (Barnes et al., 2

2013; Fiorentini et al., 2011; 2010). The Alexo sample was first isotopically defined by 3

Bekker et al. (2009). A 2 cm x 3 cm x 0.5 cm pyrrhotite-rich layer of peridotite was trimmed 4

from a larger hand sample. Alexo is mounted as 1 mm diameter pucks from this sample for 5

SIMS analysis. 6

7

3.4 Pentlandite – VMSO 8

The VMSO pentlandite is from an amphibolite-facies massive sulfide lens from the Victor 9

South shoot in the komatiite hosted nickel-sulfide Long-Victor deposit of the Kambalda 10

camp. The Kambalda camp is situated in the ca. 2.7 Ga Kalgoorlie Terrane in the Yilgarn 11

Craton of Western Australia (Barnes et al., 2013 and references therein). At Kambalda, 12

nickel-sulfide deposits are associated with thick channelized komatiite flow units in 13

stratigraphic contact with a thick pile of pillowed and massive tholeiitic basalts (Lunnon 14

Basalt Formation), locally overlain by sulfidic metasedimentary rocks (e.g., Lesher and 15

Groves, 1986). Subsequent folding has rotated the komatiitic channels from 30° to a vertical 16

dip to the east and 10° plunge to the south as well as resulting in the localised remobilisation 17

of some of the original sulfides into new structurally controlled positions (Stone et al., 2005). 18

The VMSO sample was first isotopically defined by Bekker et al. (2009). It comprises 0.1–19

0.5 mm magmatic pentlandite grains interlayered at the cm-scale with pyrrhotite and pyrite 20

from the basal sulfide layer of the Silver Lake Member of the Kambalda Formation. A 1 cm 21

wide pentlandite rich layer trimmed from the sample. From this layer, VMSO is mounted as 1 22

mm diameter pucks from the host massive sulfide lens for SIMS analysis. 23

24

4. RESULTS 25

26

The following section presents chemical and isotopic results pertaining to standard 27

development work for four new sulfide reference materials: pyrite (Sierra), chalcopyrite 28

(Nifty-b), pyrrhotite (Alexo), and pentlandite (VMSO). A summary of mineral chemical 29

composition collected by WDS for each reference material is presented in Table 1. The main 30

sulfide forming elements and common trace elements are presented in the Supplementary 31

Materials. Figure 2 shows select elemental maps (by WDS) of reference material. 32

33

Table 1: Summary of the chemical composition determined by wavelength dispersive 34

spectrometry. Uncertainty is two standard deviations (2SD). 35

Standard Mineral Formula Chemical composition (wt.%)

Fe (2SD) Other (2SD) Co (2SD) S (2SD)

Sierra Pyrite FeS2 53.3

(0.5) - -

46.7

(0.4)

Nifty-b Chalcopyrite CuFeS2 34.5

(1.4)

Cu: 29.8

(1.2) -

34.7

(0.5)

12

Alexo Pyrrhotite Fe0.90S 60.5

(1.2) - -

38.6

(0.4)

VMSO Pentlandite Fe4.1Ni4.8Co0.1S8 29.4

(0.8)

Ni: 35.8

(0.9)

0.3

(0.1)

32.9

(0.4)

1

2

Figure 2: Compositional maps of: A) Sierra pyrite (Fe wt.%), B) Nifty-b chalcopyrite (Cu wt.%), C) Alexo 3 pyrrhotite (Fe wt.%), D) VMSO pentlandite (Ni wt.%). Wavelength dispersive spectrometry maps are collected 4 by EPMA. 5

6

Table 2 shows the accumulated SIMS isotopic data to define analytical repeatability of the 7

four reference materials. Uncertainty reported in Table 2 and the text is twice the standard 8

deviation of the mean. A compilation of in situ sulfur isotopic measurements for reference 9

materials and unknown samples is presented in Supplementary Material. 10

11

Table 2: In situ SIMS sample repeatability for four presented reference materials and their 12

corresponding uncertainty (reported as twice the standard deviation of the mean). “n” is the 13

number of measurements. 14

15

16

13

1

2

3

4

5

6

7

8

9

10

Table 3 and text presents the weighted mean bulk sulfur isotope compositions (δ33SV-CDT, 11

δ34SV-CDT, δ36SV-CDT) measured by fluorination gas-source mass spectrometry (from multiple 12

aliquots of each reference material) and associated uncertainty, defined as two standard 13

deviations of the mean. All sulfur isotope results presented in the text are reported on the V-14

CDT scale. Figure 3 demonstrates δ34S isotope homogeneity of the reference materials. To 15

demonstrate that the reference material was homogeneous within the analytical uncertainty of 16

the fluorination gas-source mass spectrometry, an MSWD test was completed as detailed in 17

Wing and Farquhar (2015). Briefly, a population was considered homogeneous if all values 18

within a given sample population were consistent with a single mean value and returned 19

within a 95% confidence interval. Because the analytical error is well defined, MSWD is 20

used to estimate an independent “goodness of fit” to demonstrate separately that the reference 21

material is homogeneous and consistent within a single mean value. Bulk fluorination data 22

for four new reference materials are presented in the Supplementary Material. 23

24

Table 3: Weighted mean bulk sulfur isotope compositions (δ33SV-CDT, δ34SV-CDT, δ36SV-CDT, 25

Δ33SV-CDT, Δ36SV-CDT) as measured by fluorination gas-source mass spectrometry. 2SD is two 26

standard deviations of the mean. SE is the standard error associated with the mean, and 27

MSWD is an independent goodness-of-fit check after Wing and Farquhar (2015). “n” is the 28

number of independent extractions and measurements of separate aliquots. 29

30

31

32

Standard SIMS sample repeatability (‰)

δ33S

(2SD)

δ34S

(2SD)

δ36S

(2SD)

Δ33S

(2SD)

Δ36S

(2SD)

Sierra

(py)

0.15

(n=1417)

0.25

(n=1417)

0.90

(n=861)

0.08

(n=1417)

0.77

(n=861)

Nifty-b

(ccp)

0.14

(n=149)

0.23

(n=149)

0.63

(n=87)

0.08

(n=150)

0.48

(n=88)

Alexo

(po)

0.17

(n=340)

0.30

(n=340)

0.74

(n=199)

0.11

(n=340)

0.52

(n=199)

VMSO

(pn)

0.21

(n=246)

0.33

(n=246)

0.90

(n=105)

0.12

(n=246)

0.72

(n=105)

Ref. Bulk measurement (‰)

n δ33S 2SD 1SE MSWD δ34S 2SD 1SE MSWD δ36S 2SD 1SE MSWD Δ33S

(2SD)

Δ36S

(2SD)

Sierra

(py) 12 1.09 0.15 0.02 0.99 2.17 0.28 0.04 0.89 3.96 0.60 0.09 1.00

-0.02

(0.01)

-0.18

(0.15)

Nifty-b

(ccp) 5 -1.78 0.21 0.03 2.08 -3.58 0.44 0.07 2.22 -7.15 0.63 0.13 1.11

-0.06

(0.03)

-0.36

(0.45)

Alexo

(po) 9 1.73 0.20 0.03 1.87 5.23 0.40 0.05 1.77 10.98 0.59 0.10 0.95

-0.96

(0.04)

1.02

(0.27)

VMSO

(pn) 11 1.66 0.24 0.02 2.67 3.22 0.51 0.05 2.89 6.37 0.83 0.09 1.92

0.00

(0.02)

0.24

(0.35)

14

1 Figure 3: Probability density plots for δ34S for normalized reference material to demonstrate the low degree of 2 uncertainty, normal distribution and lack of outliers. 3 4

4.1 Pyrite – Sierra 5

The chemical composition of Sierra, determined by EPMA (n=78 from five fragments), 6

reveals that it is composed of 46.7 ± 0.5 wt.% Fe and 53.3 ± 0.4 wt.% S, with trace amount of 7

Co (~400 ppm). WDS mapping further demonstrates that Sierra grain fragments are 8

chemically homogeneous (Figure 2a) and free of inclusions. Twelve different fragments of 9

the 2 cm3 piece known as Sierra were analysed by fluorination gas-source mass spectrometry 10

and yielded weighted averaged values of: δ33S = 1.09 ± 0.15‰, δ34S = 2.17 ± 0.28‰, and 11

δ36S = 3.96 ± 0.60‰. Values for Δ33S and Δ36S are calculated as -0.02 ± 0.01‰ and 0.18 ± 12

0.15‰, respectively, indicating that Sierra does not contain a MIF signature. Sierra has been 13

measured as a primary reference for triple sulfur isotopes 1417 times amongst which 861 14

times included quadruple sulfur isotopes over 59 analytical sessions (see Figure 3). 15

Reproducibility of SIMS analyses on Sierra was δ33S = 0.15‰, δ34S = 0.24‰, and δ36S = 16

1.04‰. Reproducibility on Δ33S = 0.08‰ and Δ36S = 0.77‰. Different fragments of the 17

much larger pyrite, known as ‘Sonora’, have been described by Farquhar et al. (2013), Evans 18

et al. (2014) and Wacey et al. (2011). The Sonora fragments have slightly lighter values of 19

δ33S (0.83‰), δ34S (1.61‰), and δ36S (3.25‰) than those we determine here for Sierra. 20

21

4.2 Chalcopyrite – Nifty-b 22

15

WDS analyses (n = 352) reveal that Nifty-b chalcopyrite is composed of 34.5 ± 1.4 wt.% Cu, 1

29.8 ± 1.2 wt.% Fe, and 34.7 ± 0.5 wt.% S, with trace amounts of Zn (~300 ppm). WDS 2

mapping demonstrates that Nifty-b is homogeneous and inclusion-free but contains fractures 3

that can be avoided during isotope measurements (Figure 2b). Five different grain fragments 4

of Nifty-b were analysed by fluorination gas-source mass spectrometry and yielded a 5

weighted average of: δ33S = -1.78 ± 0.21‰, δ34S = -3.58 ± 0.44‰, and δ36S = -7.15 ± 0.63‰. 6

Values for Δ33S and Δ36S are calculated as 0.06 ± 0.03‰ and 0.36 ± 0.50‰, respectively (the 7

large error on Δ36S prevents observation of MIF in this isotopic system). Nifty-b has a small 8

but consistent MIF signature. Nifty-b has been measured as a primary chalcopyrite reference 9

for triple sulfur isotopes 149 times, and for quadruple sulfur isotopes 87 times in 15 10

analytical sessions over the course of three years. Average measurement repeatability of 11

SIMS analyses on Nifty-b was δ33S = 0.14‰, δ34S = 0.23‰, and δ36S = 0.63‰. 12

Reproducibility on Δ33S = 0.08‰ and Δ36S = 0.48‰. 13

14

4.3 Pyrrhotite – Alexo 15

Alexo pyrrhotite has very fine exsolution lamellae, <10 µm in width, which are 16

heterogeneously dispersed. WDS analyses of Alexo (n = 61) reveal that it is composed of 17

60.5 ± 1.2 wt.% Fe and 38.6 ± 0.4 wt.% S. Exsolution lamellae are composed of cobalt-18

bearing pentlandite with a composition of 32.3 ± 1.7 wt.% Fe, 32.3 ± 1.5 wt.% Ni, 1.2 ± 0.1 19

wt.% Co, and 33.1 ± 0.6 wt.% S (n = 23). These pentlandite lamellae are very fine and WDS 20

mapping demonstrates that they are unevenly distributed throughout the grain (see Figure 2c), 21

making them difficult to observe during SIMS analysis. Alexo pyrrhotite contains inclusions 22

of magnetite (50 µm) and conjugate fracture sets; both features can be readily avoided during 23

SIMS analysis. 24

25

Nine different grains of Alexo were analysed by fluorination gas-source mass spectrometry 26

and yielded a weighted average of: δ33S = 1.73 ± 0.20‰, δ34S = 5.23 ± 0.40‰, and δ36S = 27

10.98 ± 0.59‰. Uncertainty is reported as the standard error of the mean. Values for Δ33S 28

and Δ36S are calculated as -0.96 ± 0.04‰ and 1.02 ± 0.27‰, respectively, indicating that 29

Alexo has a MIF signature. The values represent the bulk composition of Alexo, 30

incorporating both pyrrhotite and the pentlandite exsolution lamellae. Alexo has been 31

measured as a primary pyrrhotite reference for triple sulfur isotopes 340 times, and for 32

quadruple sulfur isotopes 199 times in 27 analytical sessions (see Figure 5). Reproducibility 33

of SIMS analyses on Alexo was δ33S = 0.17‰, δ34S = 0.30‰, and δ36S = 0.74‰. 34

Reproducibility on Δ33S = 0.11‰ and Δ36S = 0.52‰. 35

36

4.3.1 The effect of pentlandite exsolution in Alexo pyrrhotite 37

Unmixing of the solid solution pyrrhotite-pentlandite (Fe,Ni)1-xS at temperatures below ~600 38

ºC via exsolution of pentlandite is a common and ubiquitous petrochemical feature of 39

magmatic nickel-sulfide ore deposits (e.g., Kelly and Vaughn, 1983; Naldrett et al., 1967). To 40

unravel and decipher the sulfur isotopic record hosted by pyrrhotite grains from magmatic 41

deposits, it is important to understand the effect of potential isotope fractionation between the 42

two phases of the Alexo pyrrhotite reference material (also derived from a magmatic ore 43

deposit). The sulfur isotopic composition of Alexo, determined by fluorination gas-source 44

16

mass spectrometry, itself incorporates a small amount (1–5%) of pentlandite exsolution. This 1

investigation is crucial because: 1) lamellae may be fast pathways of isotopic exchange in an 2

open system, 2) instrumental mass fractionation may be affected by the incorporation of more 3

than one phase (see Hervig et al., 2002), and 3) natural isotopic fractionation of pentlandite-4

pyrrhotite may be reflected in the bulk analysis and not represented in the SIMS analysis 5

(when avoiding exsolution lamellae). 6

7

Because lamellae are so fine and difficult to observe even with adequate BSE imaging, it is 8

impossible to solely analyse exsolution lamellae (Figure 4). Therefore, to evaluate the effect 9

on sulfur isotope composition of the presence of exsolution lamellae within Alexo pyrrhotite 10

on sulfur isotope composition, 60 in situ analyses were carried out in two groupings: 1) 30 11

analyses of areas with an abundance of exsolution lamellae, and 2) 30 analyses of areas free 12

from exsolution lamellae. Alexo grains were reimaged after SIMS work to identify which 13

analyses incorporated some component of exsolution lamellae (Figure 9). Group one yields 14

average δ33S, δ34S, δ36S values equal to 1.75 ± 0.20‰, 5.27 ± 0.43‰, and 11.12 ± 0.98‰, 15

respectively. Group two yields average δ33S, δ34S, δ36S values equal to 1.70 ± 0.22‰, 5.18 ± 16

0.44‰, and 10.83 ± 0.98‰, respectively. Analyses for the entire run were normalised to the 17

bulk value of Alexo, a value that incorporates some component of pentlandite exsolution, but 18

that returns a low MSWD (1.8) across nine separate grain extractions and analyses. These 19

results demonstrate that exsolution of pentlandite within Alexo pyrrhotite are minimal 20

enough to be masked by the measurement repeatability and so do not impact sulfur isotope 21

measurements. Therefore, Alexo is an adequate reference material for the determination of 22

sulfur isotope composition of pyrrhotite. 23

24

25

17

Figure 4: Example of compositionally homogeneous area of Alexo pyrrhotite compared to areas of significant 1 pentlandite exsolution. Raster spots show examples of assessment of two different areas of the grain to 2 determine effect on instrumental mass fractionation. 3

4

4.4 Pentlandite – VMSO 5

WDS analyses demonstrate that VMSO pentlandite is composed of 29.4 ± 0.8 wt.% Fe, 35.8 6

± 0.9 wt.% Ni, 0.3 ± 0.1 wt.% Co, and 32.9 ± 0.4 wt.% S (n = 45), with trace amounts of Sb 7

(~150 ppm), Se (~600 ppm) and Cu (~500 ppm). VMSO pentlandite contains small (~100 8

µm) inclusions of magnetite and is intergrown with pyrrhotite and pyrite. WDS mapping 9

demonstrates that the pentlandite portion of VMSO is homogeneous but that fractures and 10

sulfide intergrowths should be avoided (Figure 4d). Analysis of the intergrowths reveal 11

compositions of 59.5 ± 0.6 wt.% Fe and 39.3 ± 0.1 wt.% S, for pyrrhotite (n = 13) and 45.6 ± 12

0.2 wt.% Fe, 0.9 ± 0.0 wt.% Co, and 53.5 ± 0.2 wt.% S, for pyrite (n = 11). The pyrrhotite 13

and pyrite components and oxide inclusions are readily identified in BSE imaging and can be 14

easily avoided during SIMS analysis. 15

16

Eleven different grains of the pentlandite portion of VMSO were analysed by fluorination 17

gas-source mass spectrometry and yielded a weighted average of: δ33S = 1.66 ± 0.24‰, δ34S 18

= 3.22 ± 0.51‰, and δ36S = 6.37 ± 0.83‰. Uncertainty is reported as the standard error of the 19

mean. Values for Δ33S and Δ36S are calculated as 0.00 ± 0.02‰ and 0.24 ± 0.35‰, 20

respectively, indicating that VMSO does not have a MIF signature. The elevated MSWD on 21

sulfur isotope measurements of VMSO is discussed in section 5.1.3. VMSO has been 22

measured as a primary pentlandite reference for triple sulfur isotopes 255 times, and for 23

quadruple sulfur isotopes 98 times in 22 analytical sessions (see Figure 5). Reproducibility of 24

SIMS analysis on VMSO was δ33S = 0.21‰, δ34S = 0.33‰, and δ36S = 0.90‰. 25

Reproducibility on Δ33S = 0.12‰ and Δ36S = 0.72‰. 26

27

4.4.1 Investigation into the homogeneity of VMSO pentlandite 28

Bulk measurements of VMSO pentlandite yield larger uncertainties and greater MSWD 29

values (δ33S = 1.66 ± 0.23‰ (MSWD 2.67), δ34S = 3.22 ± 0.51‰ (MSWD 2.89), and δ36S = 30

6.37 ± 0.83‰ (MSWD 1.92)) than other sulfide reference material, indicating that it is a 31

more isotopically heterogeneous material. Pentlandite does not exist naturally as large single 32

crystals; therefore, this reference material has a certain level of isotopic heterogeneity 33

attributed to micro-scale intergrowths of pentlandite and pyrrhotite within the reference 34

material, a feature typical of sulfide ore in magmatic deposits. Although these two sulfide 35

phases are cogenetic, slightly different isotopic compositions between the two phases are 36

expected due to natural equilibrium isotopic fractionation. The <mm-scale intergrowths are 37

large enough to be easily avoidable during SIMS analysis; however, the incorporation of a 38

small component of pyrrhotite during extraction of (relatively) large volumes of sulfur for 39

bulk analysis is unavoidable (see pyrrhotite-pentlandite intergrowth in Figure 2d). 40

41

To evaluate the sulfur isotope composition of the pyrrhotite portion of VMSO, seven in situ 42

SIMS analyses of the pyrrhotite intergrowths were normalized using Alexo pyrrhotite. Sulfur 43

isotope analysis of pyrrhotite intergrowths yielded values of δ33S = 2.07 ± 0.26‰, δ34S = 44

18

3.86 ± 0.18‰, and δ36S = 8.09 ± 0.26‰. These values are slightly elevated to the determined 1

isotopic composition of VMSO pentlandite but within the envelope of uncertainty. Therefore, 2

the uncertainty on the bulk values mask the incorporation of a component of pyrrhotite 3

during the extraction process. We determine VMSO reference material to be adequate for in 4

situ sulfur isotope analysis of pentlandite but to be used with the expectation of propagating 5

poorer analytical precision. 6

7

5. DISCUSSION 8

Mineral systems are natural laboratories that provide invaluable information on the flux of 9

metals and fluids that are cycled among the different reservoirs of our planet. Most of this 10

information is stored as chemical and isotopic signatures, which are preserved in the 11

magmatic and hydrothermal sulfide phases that comprise the mineralization hosted in 12

different mineral systems. However, the interpretation of the information stored in sulfides 13

requires care, as the minero-chemical features of sulfides are prone to be reset over a wide 14

range of physical conditions. The resulting complexity recorded in the chemical and isotopic 15

nature of any given sulfide assemblage may be cryptic to bulk rock analysis, but can be 16

resolved with an in situ analytical approach, which allows the investigator to carefully select 17

specific grains or crystal fractions and avoid others. 18

19

5.1 Selecting and validating sulfide reference material 20

In the case of selecting reference material for pyrite and chalcopyrite, small pieces (~2 cm3) 21

of much larger crystals from low grade hydrothermal deposits yield a much higher degree of 22

chemical and isotopic homogeneity over small individual crystals that make up sulfide-rich 23

layers. Here, we demonstrate this fact by comparing the Sierra standard (a fragment of a large 24

hydrothermal crystal) to other available pyrite sulfur isotope standards that occur as small 25

individual grains within a stratigraphic horizon. 26

27

Grains of Isua 248474 pyrite (kindly supplied by Martin Whitehouse) from the Isua 28

greenstone belt, SW Greenland (Whitehouse, 2013) were used as secondary standards to 29

assess the adequacy of Sierra as a reference material. Isua 248474 has reported bulk sulfur 30

isotope values of δ33S = 4.33 ± 0.24‰ (1σ), δ34S = 1.99 ± 0.18‰ (1σ) (Baublys et al., 2004). 31

A total of 216 SIMS analyses (in 19 analytical sessions) have been completed on Isua 248474 32

pyrite as an unknown using Sierra as the primary reference material and yield an average of 33

δ33S = 4.45 ± 0.66‰, δ34S = 2.60 ± 0.86‰, and δ36S = 2.74 ± 1.60‰ and a calculated MIF 34

signature: Δ33S = 3.11 ± 0.33‰ and Δ36S = -2.38 ± 0.81‰. Variability within the Isua 35

248474 pyrite standard has been observed by Whitehouse (2013) who reports some 36

dispersion to lower values for δ34S but yields an average δ34S that is higher but within error of 37

the accepted bulk measurement of Baublys et al. (2004). Here, our Isua 248474 38

measurements normalised to Sierra have a high uncertainty, consistent with the isotopic 39

variability observed by Whitehouse (2013). 40

41

A grain of UW Balmat pyrite (kindly supplied by John Craven) from the Adirondack 42

Mountains, New York was also assessed. UW Balmat pyrite was originally described by 43

Crowe et al. (1990) who reported δ34S equal to 14.63 ± 0.38‰ (2SD). Further to that, 44

19

Graham and Valley (1992) describe the UW Balmat pyrite as only being isotopically 1

homogeneous at the ~1.0‰ level. Pyrite from a slightly different stratigraphic section of the 2

deposit (UWPy-1) does not show any MIF anomaly (Williford et al., 2011). Two analytical 3

sessions for triple isotopes give 31 analyses of UW Balmat pyrite with an average δ34S of 4

16.23 ± 0.18‰ (2SD). Our values for δ33S equal 8.30 ± 0.12‰ and yield a Δ33S of -0.03‰, 5

confirming the lack of a MIF signature. We attribute the variability in both the Isua 248474 6

and UW Balmat pyrite standards to being separate grains distributed between laboratories. 7

8

In the development of a chalcopyrite standard, four separate materials were tested for isotopic 9

homogeneity. Two materials were fragments of large hydrothermal chalcopyrite crystals 10

which yielded a much higher degree of δ34S reproducibility than the two other materials 11

derived from massive sulfide horizons in magmatic systems. The pyrite and chalcopyrite 12

results demonstrate that small fragments of large hydrothermal sulfide crystals are more 13

isotopically homogeneous and more suitable standards than individual sulfide grains in 14

sulfide-rich layers. 15

16

In the case of pyrrhotite and pentlandite, these minerals do not commonly occur as large 17

hydrothermal crystals and so the most homogeneous reference material were selected from 18

massive sulfide layers from magmatic ore deposits. Preliminary investigation into how 19

chemical heterogeneity (i.e., micro-exsolution and intergrowths between the two phases) 20

affects isotopic homogeneity in sections 4.3 and 4.4, demonstrate that Alexo and VMSO 21

reference materials are suitable for the determination of sulfur isotope composition of 22

pyrrhotite and pentlandite and ideal for the investigation of magmatic deposits (in which 23

unknown sulfides will have similar textures). However, for comparison, further effort to 24

development a hydrothermal pyrrhotite standard is warranted. 25

26

We have presented chemical and isotopic compositions for a suite of chemically and 27

isotopically homogeneous sulfide reference materials that are representative of the sulfide 28

mineralogy of magmatic and hydrothermal mineral systems. Figure 5 presents plots of: 1) 29

δ33SV-CDT vs. δ34SV-CDT, and 2) Δ33S vs. δ34SV-CDT for all reference material analysed in this 30

study. Two of the four sulfide standards (pyrrhotite and chalcopyrite) contain a MIF 31

component, a feature that is recommended to confidently assess the accuracy of Δ33S and 32

Δ36S measurements (Whitehouse, 2013). SIMS measurements on the sulfur isotope dataset 33

demonstrate that the reference materials are adequately isotopically homogeneous and return 34

repeatability on the order of uncertainty on the bulk measurements. Furthermore, bulk 35

measurements deem the reference material homogeneous by the MSWD test as detailed in 36

Wing and Farquhar (2015). 37

38

20

1

Figure 5: Plots of: (a) δ33SV-CDT vs. δ34SV-CDT and (b) Δ33SV-CDT vs. δ34SV-CDT for all sulfide reference material 2

(Sierra pyrite, Nifty-b chalcopyrite, Alexo pyrrhotite, and VMSO pentlandite) measured by SIMS in this study. 3 Error crosses are the overall uncertainty (at the 2SD level) based on propagating the individual analysis 4 repeatability with the reproducibility for the relevant analytical session. The Sierra pyrite is used as the primary 5 reference for measuring previously characterised pyrite reference material Isua 248474 and UW Balmat pyrite 6 (see Whitehouse, 2013; Graham and Valley, 1992). 7

8

5.2 In situ multiple sulfur isotope analysis of a nickel-sulfide deposit, Kambalda 9

The in situ SIMS multiple sulfur isotope data from this study are used to constrain the nature 10

of the ore-forming process in the Moran shoot, within the Kambalda camp of Western 11

Australia. The results are compared with existing data from the Victor South shoot along the 12

Victor channel, where the VMSO reference material was collected and analysed, in order to 13

reflect on the sulfur isotope architecture of the various channels exposed in the Long-Victor 14

deposit. The integrated dataset also permits discussion of the large-scale setting of the 15

Kambalda camp in relation to other world-class nickel-sulfide deposits in the Kalgoorlie 16

21

Terrane of the Yilgarn Craton, which hosts the largest resources and reserves of nickel-1

sulfide mineralisation associated with komatiites (Barnes and Fiorentini, 2012). We utilise 2

the case study to validate the presented reference material, assess the natural fractionation 3

factor of δ34S between pyrrhotite and pentlandite, and assess whether crystallographic 4

orientation of pentlandite induces biases on δ34S and Δ33S measurements. 5

6

5.2.1 Moran shoot, Long-Victor deposit, Kambalda camp 7

The case study focussed on the identification of the multiple sulfur isotope signature of the 8

major magmatic sulfide phases (pyrrhotite, pentlandite and chalcopyrite) hosted in the 9

massive sulfide layer at the base of the Moran shoot at Kambalda, Western Australia. The 10

spatial and genetic relationship between the Victor South shoot, where VMSO reference 11

material was collected, and the Moran shoot, where the sample for the case study was 12

collected, is discussed in Barnes et al. (2013) and presented in Figure 6. Briefly, the Moran 13

shoot is hosted in the same channelized structure that hosts the larger Long shoot. This 14

channel runs roughly parallel to the Victor channel in which the Victor South shoot is hosted. 15

The Moran shoot attains 500 m along strike but is open to the south, where the boundaries of 16

the mineralised body are not defined. 17

18

19

Figure 6: Location of the Victor-Long nickel-sulfide deposit of the Kambalda camp, Western Australia after 20 Bekker et al. (2009) and references therein. The Victor-Long deposit is host to the Victor South and Moran 21 shoots, discussed in this study. 22

22

The ore body largely comprises massive and matrix sulfides dominantly composed of 1

pentlandite (38%) and pyrrhotite (58%), with minor (4%) chalcopyrite (Figure 7). Pyrrhotite 2

and pentlandite are intergrown and are in equilibrium. Chalcopyrite forms small (<40 µm) 3

interstitial grains in equilibrium with the other sulfides at pentlandite-pyrrhotite boundaries. 4

Multiple sulfur isotope data were acquired from four 5 mm in diameter pucks of ore and 5

values for δ34S, Δ33S, Δ36S are presented in Table 4. WDS data were also acquired from the 6

pentlandite and pyrrhotite to ensure similar chemical composition to the reference material. 7

8

9

Figure 7: A) Hand sample of the Moran Shoot massive sulfide horizon of the Long-Victor deposit, Kambalda 10 camp. Numbers demarcate localities drilled for bulk analysis. B) Micro-XRF map of slab of nickel-sulfide ore 11 from the Moran shoot, Long-Victor deposit, Kambalda camp. Massive sulfide is composed dominantly of 12 pentlandite (light blue) and pyrrhotite (dark blue) with minor chalcopyrite (green). The proportion of sulfides is 13 calculated from the Micro-XRF map to be 38% pentlandite (pn), 58% pyrrhotite (po), and 4% chalcopyrite 14 (ccp). C) EDS phase map showing distribution of NiS (pentlandite; yellow), FeS (pyrrhotite; red), and FeO (iron 15 oxide; green). D) EDS Cu map showing chalcopyrite in green. E) Reflected light photomicrograph of 16 pentlandite (yellow)-pyrrhotite (grey) association with small interstitial grains of chalcopyrite (orange). Area 17 delineated by black box in C. Ion probe pits are small black boxes and are 15 µm in length. 18

19

Table 4: Multiple sulfur isotope results (δ34SV-CDT, Δ33SV-CDT, Δ36SV-CDT) acquired by SIMS 20

using reference material presented in this manuscript to correct for instrumental mass 21

fractionation. “n” is the number of measurements. 22

Chemical

composition Sulfur isotope composition (‰)

δ34S 2SD Δ33S 2SD Δ36S 2SD n

pentlandite Fe4.1Ni4.8Co0.1S8 2.91 0.35 0.25 0.10 -0.03 0.40 73

pyrrhotite Fe0.90S 3.90 0.20 0.26 0.13 -0.01 0.24 19

chalcopyrite CuFeS2 3.10 0.54 0.16 0.08 -0.04 0.21 13

23

The in situ SIMS data presented in Table 4 compare well with 10 aliquots of bulk sulfide 24

material from the same ore shoot, which returned δ34SV-CDT, Δ33SV-CDT, Δ36SV-CDT equal to 25

23

3.10 ± 0.70‰ (2SD), 0.24 ± 0.04‰ (2SD), and -0.21 ± 0.10‰ (2SD). The large uncertainty 1

on δ34S in the bulk data demonstrates that the yielded sulfur isotope results represent an 2

average isotopic composition of the three sulfide phases (as depicted by the SIMS δ34S 3

results). 4

5

The case study is aimed at emphasizing the significance of a multi-sulfide phase in situ 6

analytical approach in the determination of the multiple sulfur isotope signature of Archean 7

komatiite-hosted nickel-sulfide deposits. Results indicate that the magmatic sulfide 8

assemblage of the massive sulfide horizon hosted in the Moran shoot of the Long-Victor 9

deposit in the Kambalda region of Western Australia comprises pentlandite, pyrrhotite and 10

chalcopyrite. These phases are in isotopic equilibrium and display similar values for Δ33S = 11

~+0.2‰ and Δ36S = ~−0.02‰. However, the three sulfide phases show variable δ34S values 12

(δ34Spentlandite = ~2.9‰, δ34Schalcopyrite = ~3.1‰, and δ34Spyrrhotite = ~3.9‰), which are indicative 13

of natural isotope fractionation. Natural δ34S fractionation among the sulfides that make up a 14

multi-sulfide phase ore assemblage cannot be easily resolved through bulk rock analysis (as 15

presented in Figure 8). This in turn may lead to misinterpretations of subtle sulfur isotope 16

variations. In situ analysis also has the potential to highlight the presence of secondary 17

sulfides. Therefore, the in situ approach provides the user the ability to resolve ambiguities 18

that may arise through whole-rock isotope analysis. 19

20

21

Figure 8: Probability density plots (bin size 10) of δ34S and Δ33S pyrrhotite, pentlandite, and chalcopyrite results 22 for the Moran shoot of the Long-Victor deposit, Kambalda. A representative subset of pentlandite results are 23 chosen. All three sulfide phases yield the same Δ33S values (+0.2‰) demonstrating they are likely cogenetic and 24 from the same source. The variable δ34S are a result of natural isotopic fractionation between the three sulfide 25 phases (of which pentlandite and pyrrhotite compose 96% of the ore body). Variability in δ34S chalcopyrite is a 26 product of few analyses (13) of small interstitial grains. Whole rock bulk fluorination values (from 10 aliquots) 27 depicted by dashed line and uncertainty (2SD) by grey box. 28

29

5.2.2 Source of sulfur to trigger sulfide saturation 30

24

The multiple sulfur isotope data collected from the Moran shoot is assessed in conjunction 1

with the multiple sulfur isotope signature of VMSO, which was collected from the Victor 2

South shoot. Bekker et al. (2009) originally measured the multiple sulfur isotope signature of 3

the ore sample from which the VMSO pentlandite standard was made. The bulk rock 4

fluorination data indicated that the massive sulfide in the Victor South shoot did not display 5

any evidence of MIF. This result was not discussed in detail in Bekker et al. (2009), but it 6

clearly portrayed Kambalda as being radically different from the giant komatiite-hosted 7

deposits in the northern part of the Kalgoorlie Terrane, which display a very pronounced 8

negative MIF signature (up to -0.7‰). The bulk rock isotope signature of the VMSO ore 9

sample is reflected in the lack of MIF recorded in the VMSO pentlandite standard presented 10

in this study. 11

12

In the Long-Victor deposit, the relationship between the Victor South and Moran shoots is 13

described in Barnes et al. (2013). For the sake of the argument here, the most important 14

message to be taken away from the study of Barnes et al. (2013) is that different shoots, 15

which originated from the emplacement of komatiite magma along spatially close but distinct 16

volcanic channels, may have resulted from the equilibration of variable proportions of sulfide 17

and silicate liquids. Evidence of this variable proportion, which is expressed as R Factor 18

(Campbell and Naldrett, 1979), is not only recorded in the highly variable metal tenor of the 19

ore shoots and associated variable degrees of chalcophile element enrichment and depletion 20

in the host komatiite channelized units, but it is also preserved in the sulfur isotopic 21

architecture of the deposit, as discussed in Lesher and Burnham (2001) and Fiorentini et al. 22

(2012a). This should serve as a word of caution for any exploration targeting tool based on 23

sulfur isotope analysis to track crustal assimilation as a proxy for magma fertility. 24

25

Evidence of a positive MIF anomaly (Δ33S = 0.2‰) recorded in three sulfide phases of the 26

Moran shoot suggests thermo-mechanical assimilation of sulfidic shales (Δ33S = up to 0.8‰; 27

Bekker et al., 2009) during komatiite emplacement. The magnitude of MIF as expressed by 28

Δ33S is small and so to relate the sulfur isotope signature to photochemical reactions related 29

to an oxygen-poor atmosphere (rather than mass dependent processes as described by 30

Johnston, 2011), we evaluate the Δ36S/Δ33S relationship. The Δ36S/Δ33S ratio ranges from 31

−0.04 to −0.88, extending the known Archean ratio beyond −1 to −2 of Johnston (2011). The 32

ratio is significantly different to the predicted range of mass dependent fractionation 33

processes (Δ36S/Δ33S = −9 to −7; Ono et al., 2006b; Johnston et al., 2007). 34

35

The lack of any MIF anomaly in the Victor South shoot suggests higher R factor values for 36

the channel, which was effectively flushed with higher volumes of mantle-derived silicate 37

magma carrying no MIF anomaly, thus diluting any evidence of crustal MIF. However, it is 38

crucial to state that in magmatic systems, near-zero δ34S or Δ33S values permit a mantle 39

source (δ34S = −1.28 ± 0.33‰; Δ33S = 0.00‰; Labidi et al., 2013) but do not prove it if the 40

full complexity of homogenization processes is considered. However, non-zero δ34S or Δ33S 41

values prove a non-mantle source and can be further confirmed by the Δ36S/Δ33S ratio 42

(Farquhar et al., 2007; Johnston, 2011). The data from two different shoots at Kambalda 43

highlight the variation that characterizes ore-forming processes in magmatic systems. This 44

25

variation is translated downstream in the complexity of the interpretation of chemical and 1

isotopic data that are used as proxies to target mineralisation in exploration (Fiorentini et al., 2

2012a). 3

4

In addition to the deposit-scale observations discussed above, the in situ multiple sulfur 5

isotope data presented in this study also provide key insights into the geodynamic setting 6

where the Kambalda komatiites were emplaced. Unlike the giant dunite-hosted komatiite 7

systems that thermo-mechanically assimilated volcanogenic massive sulfides proximal to 8

vents in the Agnew-Wiluna greenstone belt (cf. Fiorentini et al., 2012b), the integrated results 9

from both the Moran (case study) and Victor South (VMSO sample) shoots indicate that 10

Kambalda ores assimilated abyssal sulfidic shales. This inference is based on the very 11

different MIF signatures that are observed: the dunite-hosted systems in the northern 12

Kalgoorlie Terrane (Fiorentini et al 2012b) and elsewhere globally (e.g., Heggie et al., 2012a, 13

b) commonly display negative Δ33S values (up to −1.0‰), whereas Kambalda-style 14

channelized systems either display no MIF or positive (up to +1.1‰) signatures. These 15

variable inter-shoot isotopic results at Kambalda are reflected in the variable platinum group 16

element signature of the various channelized systems (Barnes et al., 2013). At the broader 17

scale, these data support the hypothesis that dunite-hosted systems are generally 18

characterized by lower R factors, where evidence of crustal contamination from proximal 19

volcanogenic massive sulfides is preserved, in contrast to Kambalda-type deposits where 20

evidence of crustal contamination by sulfidic shales is partially to completely erased due to 21

extreme magma flux along channelized environments. 22

23

5.2.3 In situ data to identify natural fractionation between pentlandite-pyrrhotite 24

The complex sulfide mineralogy hosted in any mineralized sample may document various 25

stages of the ore forming process. However, even when sulfide phases are cogenetic and in 26

equilibrium, natural isotopic fractionation between sulfide pairs may occur due to crystal 27

structure, temperature of formation, and chemical composition (e.g., O’Neil, 1986). For 28

instance, the speciation of the sulfur source and the mass of a metal speciation can affect the 29

preference for 34S incorporation in one sulfide over another during precipitation (Bachinski, 30

1969; Seal, 2006 and references therein). This means that the isotopic signature of a single 31

sulfide phase hosted in a sulfide layer may not be representative of the true isotopic 32

composition of that very horizon. 33

34

Although, fractionation of sulfur isotopes between cogenetic sulfide minerals has been 35

investigated by Bachinski (1969), Li and Lui (2006), Ohmoto and Rye (1979), Seal (2006), 36

among others, the natural isotope fractionation of sulfur isotopes between pentlandite and 37

pyrrhotite has not yet been quantified; however it is an important element in the study of 38

magmatic ore deposits. In the case study focussed on the Moran shoot of the Long-Victor 39

deposit, if we interpret pyrrhotite and pentlandite to be in equilibrium we conclude that 40

pyrrhotite has a natural affinity for 34S over pentlandite. Our results that demonstrate 41

enrichment in δ34S in pyrrhotite over chalcopyrite are in agreement with Li and Lui (2006). 42

This same trend is also described for VMSO pentlandite above (also derived from a 43

26

magmatic ore deposit) and has been observed in other magmatic ore deposits (LaFlamme, 1

unpublished data). 2

3

Multiple sulfur isotope analysis of massive sulfide from the Moran shoot at Long-Victor 4

deposit yields δ34S pentlandite = 2.91 ± 0.35‰ (n=73) and δ34S pyrrhotite = 3.90 ± 0.20‰ 5

(n=19). These two phases make up >96% of sulfides in this ore body and are within error of 6

the bulk fluorination results (δ34S = 3.10 ± 0.70‰; n=10). Multiple sulfur isotope analysis of 7

the massive sulfide horizon at the Victor South shoot (from which VMSO reference material 8

is derived) yields δ34S pentlandite (measured as bulk fluorination value and incorporates a 9

minor component of unavoidable pyrrhotite) = 3.22 ± 0.51‰ (n=11) and δ34S pyrrhotite 10

(normalized to Alexo pyrrhotite) = 3.86 ± 0.18‰ (n=7). This is a circular problem because it 11

is impossible to get a true value for pentlandite unaccompanied by pyrrhotite for VMSO, and 12

the values for in situ analyses from the Moran shoot rely on a correction for instrumental 13

mass fractionation normalised to the VMSO reference material. Here though, we assume the 14

cogenetic sulfides from both shoots are in natural equilibrium, and so we can broadly 15

estimate δ34Spyrrhotite-pentlandite to be 0.7–1.0 ± 0.5‰. The natural equilibrium isotopic 16

fractionation between pyrrhotite and chalcopyrite determined at the Moran shoot to be δ34S = 17

~0.2‰, is in agreement with that defined by Li and Lui (2006). Therefore, in ore bodies with 18

multiple sulfide phases in equilibrium, δ34S values are a reflection of the relative proportions 19

of those phases. Therefore, care should be applied when small deviations in isotopic 20

signatures derived from bulk analyses of massive sulfide bodies are used to interpret 21

variations in the source of sulfur, volatiles and metals in ore bodies (e.g., Bekker et al., 2009; 22

Fiorentini et al., 2012b; Li et al., 2009). In fact, these variations may solely be reflecting 23

proportional differences of sulfide phases, or may homogenise different phases of 24

mineralisation with different isotopic signatures. In turn this can lead to a better 25

understanding of small spatial and temporal changes in multiple sulfur isotopes related to the 26

magmatic and/or hydrothermal environment. 27

28

5.2.4 Test for potential crystallographic orientation effects in pentlandite 29

The integration of EBSD with high-spatial resolution isotopic analyses allows the potential 30

role of crystal orientation on isotopic data quality to be assessed (Taylor et al., 2012). The 31

potential effects of grain orientation on sulfur isotope measurements in the Moran Shoot 32

pentlandite have been assessed by comparison of EBSD-derived crystal orientation 33

information with sulfur isotope data (as δ34S and Δ33S) from known ion probe analysis sites. 34

Results demonstrate that no orientation effect on δ34S and Δ33S exists for pentlandite (Figure 35

7). Inverse pole figures (of which one sample is shown in Figure 9) relate count rates to grain 36

orientation in which analyses are plotted based on the angular relationship between the mount 37

x-direction and the main crystal axes. Comparison of grain orientation data from ~50 38

pentlandite grains and 73 multiple sulfur isotopes analyses demonstrates a consistent value of 39

δ34S and Δ33S, within the envelope of single analysis uncertainty irrespective of grain 40

orientation. 41

42

Certain sulfide minerals (galena and sphalerite) have been shown to exhibit a crystal 43

orientation effect resulting in instrumental mass bias towards 34S, while others (pyrite, 44

27

pyrrhotite and chalcopyrite) do not (Kozdon et al., 2010; Kita et al., 2011). Here, our data 1

indicate that there is no resolvable crystallographic orientation control on either δ34S or Δ33S 2

due to the interaction of the beam with the pentlandite grain. This is a completely new 3

outcome as the orientation effect on beam-pentlandite interaction was not previously known. 4

5

6

Figure 9: Example of orientation effect in pentlandite analysed by electron backscatter diffraction (EBSD). 7 Pentlandite from massive sulfide lens mounted as pucks (1 puck shown here) drilled from a massive sulfide lens 8 at the Moran shoot, Kambalda. A: EDS phase map showing distribution of NiS (pentlandite; yellow), FeS 9 (pyrrhotite; red), and FeO (iron oxide; green). B: Inverse pole figure map, plotted with respect to the user-10 defined x direction. Black lines show grain boundaries with >10° misorientation. C: Inverse pole figure (lower 11 hemisphere equal area projection) for the sample x direction showing grain orientations indicated by colours of 12 pentlandite grains shown in B (n=17195 data points). D: δ34S and Δ33S from pentlandite with colour matched to 13 crystallographic orientation shown in B and C. 14

15

6. CONCLUSIONS 16

Instrumental mass fractionation is dependent on the chemical matrix of the material being 17

analysed (the so-called matrix effect). For different sulfide minerals, the matrix effect is 18

likely due to a combination of factors including varying sulfur contents, varying cation 19

compositions and different crystallographic systems, giving rise to different ionisation 20

potentials and different sputtering rates (e.g., Kozdon et al., 2010). The matrix effect between 21

conjugate pyrite-sulfide pairs is a complex problem (at least partly) related to the presence of 22

multiple and subtle instrument conditions and analytical protocols. Hence, SIMS inter-23

28

mineral fractionation factors (even within a single laboratory) to calculate sulfur isotope 1

ratios should be used with caution and only as a last resort. Multiple sulfur isotope studies 2

should also take caution against propagating δ33S and δ36S from δ34S following the mass 3

dependent fractionation relationship for reference materials in which these values have not 4

been quantified. It has been shown that Δ33S and Δ36S anomalies can occur in a range of 5

environments even outside of Archean terranes (i.e., at Proterozoic craton margins, Selvaraja 6

et al., 2016; this study (Proterozoic Nifty-b chalcopyrite reference material); LaFlamme, 7

unpublished data, at modern hydrothermal vents; Ono et al., 2006b; Farquhar et al., 2007; and 8

in other environments, Cabral et al., 2013). Therefore, for the reasons above, unknown sulfur 9

isotope analyses should be normalised to chemically and isotopically characterised reference 10

material, four sulfides of which are presented here. 11

12

At the Long-Victor nickel-sulfide deposit in the world class Kambalda nickel camp in the 13

southern Kalgoorlie Terrane of Western Australia, a case study demonstrates how precise 14

multiple sulfur isotope analyses from magmatic pentlandite, pyrrhotite and chalcopyrite can 15

better constrain genetic models related to ore-forming processes. Results indicate that 16

pentlandite, pyrrhotite and chalcopyrite are in isotopic equilibrium and yield the same value 17

for Δ33S (+0.2‰). The three magmatic phases show variable δ34S values (δ34Schalcopyrite = 18

2.9‰, δ34Spentlandite = 3.1‰, and δ34Spyrrhotite = 3.9‰), which are indicative of natural 19

fractionation. The results indicate that unlike the giant dunite-hosted komatiite systems that 20

thermo-mechanically assimilated volcanogenic massive sulfides proximal to vents and 21

display negative Δ33S values, the Kambalda ores formed in relatively distal environments 22

assimilating abyssal sulfidic shales. Careful in situ analysis is able to image the subtle 23

isotopic variability of the magmatic sulfide assemblage, which may help resolve the nature of 24

the ore-forming process. This approach may help to discriminate the magmatic sulfur isotope 25

signature from that recorded in metamorphic- and alteration-related sulfides, which may not 26

be resolved during bulk rock fluorination analysis. 27

28

7. ACKNOWLEDGEMENTS 29

This manuscript benefited greatly from suggestions made by Drs. Richard Stern, John Valley 30

and two anonymous reviewers. The authors thank Klaus Mezger for editorial handling. The 31

authors acknowledge the facilities and technical assistance of the Australian Microscopy and 32

Microanalysis Research Facility at the CMCA, UWA and also acknowledge the facilities of 33

the Advanced Resource Characterisation Facility at the Commonwealth Scientific and 34

Industrial Resource Organisation and the technical assistance of Michael Verrall. Crystal 35

LaFlamme acknowledges support from the Minerals Research Institute of Western Australia, 36

Science and Industry Endowment Fund and Geological Survey of Western Australia. Marco 37

Fiorentini acknowledges support from the Australian Research Council through Linkage 38

Project LP120100668, the Future Fellowship Scheme (FT110100241), and the ARC Centre 39

for Excellence for Core to Crust Fluid Systems (CE11E0070). David Wacey acknowledges 40

support from the European Commission and the Australian Research Council through the 41

Future Fellowship Scheme (FT140100321). The Stable Isotope Laboratory in the Earth and 42

Planetary Science department at McGill University is supported by the FQRNT through the 43

GEOTOP research centre. Early developmental work was completed by John Cliff. Materials 44

29

to test in the preliminary stages of standard development were provided by Alex Bevan at the 1

Museum of Western Australia. Two reference materials were supplied by Martin Whitehouse 2

from the Nordsim laboratory and John Craven from the Edinburgh Ion Microprobe Facility. 3

This is contribution 847 from the ARC Centre of Excellence for Core to Crust Fluid Systems. 4

http://www.ccfs.mq.edu.au. 5

6

8. LIST OF FIGURES 7

Figure 1: Calculated magnitudes of mass independent fractionation as deviations from the 8

mass dependent fractionation line (MDF) presented in δ33S vs. δ34S space. The small 9

deviations from the MDF line may represent mass independent fractionation. This highlights 10

the importance of: 1) precise and accurate multiple sulfur isotope measurements, and 2) 11

quantification of uncertainty on Δ33S. Although not expressed in this figure, quantifying Δ36S 12

is equally important. 13

14

Figure 2: Demonstration of compositional homogeneity of: A) Sierra pyrite (Fe wt.%), B) 15

Nifty-b chalcopyrite (Cu wt.%), C) Alexo pyrrhotite (Fe wt.%), D) VMSO pentlandite (Ni 16

wt.%). Wavelength dispersive spectrometry maps are collected by EPMA. 17

18

Figure 3: Probability density plots for δ34S for normalized reference material to demonstrate 19

the low degree of uncertainty, normal distribution and lack of outliers. 20

21

Figure 4: Example of compositionally homogeneous area of Alexo pyrrhotite compared to 22

areas of significant pentlandite exsolution. Raster spots show examples of assessment of two 23

different areas of the grain to determine effect on instrumental mass fractionation. 24

25

Figure 5: Plots of: (a) δ33SV-CDT vs. δ34SV-CDT and (b) Δ33S vs. δ34SV-CDT for all sulfide 26

reference material (Sierra pyrite, Nifty-b chalcopyrite, Alexo pyrrhotite, and VMSO 27

pentlandite) measured by SIMS in this study. Error crosses are the overall uncertainty (at the 28

2SD level) based on propagating the individual analysis repeatability with the reproducibility 29

for the relevant analytical session. The Sierra pyrite is used as the primary reference for 30

measuring previously characterised pyrite reference material Isua 248474 and Balmat pyrite 31

(see Whitehouse, 2013; Graham and Valley, 1992). 32

33

Figure 6: Location of the Victor-Long nickel-sulfide deposit of the Kambalda camp, Western 34

Australia. The Victor-Long deposit is host to the Victor South and Moran shoots, discussed 35

in this study. 36

37

Figure 7: A) Hand sample of the Moran Shoot massive sulfide horizon of the Long-Victor 38

deposit, Kambalda camp. B) Micro-XRF map of slab of nickel-sulfide ore from the Moran 39

shoot, Long-Victor deposit, Kambalda camp. Massive sulfide is composed dominantly of 40

pentlandite (light blue) and pyrrhotite (dark blue) with minor chalcopyrite (green). The 41

proportion of sulfides is calculated from the micro-XRF map to be 38% pentlandite, 58% 42

pyrrhotite, and 4% chalcopyrite. C) EDS phase map showing distribution of NiS (pentlandite; 43

30

yellow), FeS (pyrrhotite; red), and FeO (iron oxide; green). D) EDS Cu map showing 1

chalcopyrite in green. E) Reflected light photomicrograph of pentlandite (yellow)-pyrrhotite 2

(grey) association with small interstitial grains of chalcopyrite (orange). Ion probe pits are 3

small black boxes and are 15 µm in length. 4

5

Figure 8: Probability density plots (bin size 10) of δ34S and Δ33S pyrrhotite, pentlandite, and 6

chalcopyrite results for the Moran shoot of the Long-Victor Deposit, Kambalda. A 7

representative subset of pentlandite results are chosen. All three sulfide phases yield the same 8

Δ33S values (+0.2‰) meaning they are cogenetic and from the same source. The variable 9

δ34S are a result of natural isotopic fractionation between the three sulfide phases (of which 10

pentlandite and pyrrhotite composed >96% of the ore body). Whole rock bulk fluorination 11

values (from 10 aliquots) depicted by dashed line and uncertainty (2SD) by grey box. 12

13

Figure 9 Example of orientation effect in pentlandite analysed by electron backscatter 14

diffraction (EBSD). Pentlandite mounted as a puck drilled from a massive sulfide lens at 15

Kambalda. A: Phases of massive sulfide as defined by electron dispersive spectrometry as 16

NiS (pentlandite; light green), FeS (pyrrhotite; red), and FeO (iron oxide; dark green). B: 17

Inverse Pole Figure with respect to the x direction. Black lines demonstrate more than 10° 18

misorientation which are interpreted as grain boundaries. C: Inverse Pole Figure diagram for 19

the x direction (17195 data points) as equal area projection in the lower hemisphere. D: δ34S 20

and Δ33S from pentlandite with colour matched to crystallographic orientation demonstrating 21

no orientation effect with respect to crystallographic orientation. 22

23

9. REFERENCES 24

Alsen, N.1925. Roentgenographische Untersuchungen der Kristallstrukturen von Magnetkies, 25

Breithaupit, Pendlandit, Millerir und verwandten Verbindungen. Geologiska Foereningens i 26

Stockholm Foerhandlingar 47, 19-73. 27

Anderson, B.R., Gemmell, J.B., Berry, R. 2001. The geology of the Nifty Copper Deposit, 28

Throssel Group, Western Australia: Implications for ore genesis. Econ. Geol., 96, 1525-1565. 29

Barra, F., Valencia, V.A. 2014. Late Cretaceous porphyry copper mineralizaion in Sonora, 30

Mexico: Implications for the evolution of the Southwest North America porphyry copper 31

province. Mineral. Dep. 49, 879-884. 32

Bachinski, D.J., 1969. Bond strength and sulfur isotopic fractionation in coexisting sulfides. 33

Econ. Geol., 64, 56-65. 34

Barnes, S.J., Heggie, G.J., Fiorentini, M.L. 2013. Spatial variation in platinum group element 35

concentrations in ore-bearing komatiite at the Long-Victor deposit, Kambalda dome, Western 36

Australia: enlarging the footprint of nickel sulfide orebodies. Econ. Geol., 108, 913-933. 37

Baublys, K.A., Golding, S.D., Young, E., Kamber, B.S. 2004. Simultaneous determination of 38

δ33SV-CDT and δ34SV-CDT using masses 48, 49 and 50 on a continuous flow isotope ratio mass 39

spectrometer. Rapid Commun. Mass Spectrom., 18, 2765-2769. 40

31

Bekker, A., Barley, M.E., Fiorentini, M., Rouxel, O.J., Rumble, D., Beresford, S.W. 2009. 1

Atmospheric sulfur in Archean komatiite-hosted nickel deposits. Sci., 326, 1086−1089. 2

Bühn, B., Santos, R.V., Dardenne, M.A., de Oliveira, C.G. 2012. Mass-dependent and mass-3

independent sulfur isotope fractionation (δ34S and δ33S) from Brazilian Archean and 4

Proterozoic sulfide deposits by laser ablation multi-collector ICP-MS. Chem. Geol., 312, 5

163-176. 6

Campbell, I.H., Naldrett, A.J. 1979. The influence of silicate:sulfide ratios on the 7

geochemistry of magmatic sulfides. Econ. Geol., 74, 1503-1506. 8

Cabral, R.A., Jackson, M.G., Rose-Koga, E.F., Koga, K.T., Whitehouse, M.J., Antonelli, 9

M.A., Farquhar, J., Day, J.M.D., Hauri, E.H. 2013. Anomalous sulphur isotopes in plume 10

lavas reveal deep mantle storage of Archaean crust. Nat., 496, 490-494. 11

Canfield, D.E., Raiswell, R., Westrich, J.T., Reaves, C.M., Berner, R.A. 1986. The use of 12

chromium reduction in the analysis of reduced inorganic sulfur in sediments and shales. 13

Chem. Geol., 52, 149-155. 14

Chang, Z., Large, R.R., Maslennikov, V. 2008. Sulfur isotopes in sediment-hosted orogenic 15

gold deposits: Evidence for an early timing and a seawater sulfur source. Geol., 43, 971-974. 16

Chen, M., Campbell, I.H., Xue, Y., Tian, W., Ireland, T.R., Holden, P., Cas, R.A.F., Hayman, 17

P.C., Das, R. 2015. Multiple sulfur isotope analyses support a magmatic model for the 18

volcanogenic massive sulfide deposits of the Teutonic Bore Volcanic Complex, Yilgarn 19

Craton, Western Australia. Econ. Geol., 110, 1411-1423. 20

Craddock, P.R., Rouxel, O.J., Ball, L.A., Bach, W. 2008. Sulfur isotope measurement of 21

sulfate and sulfide by high-resolution MC-ICP-MS. Chem. Geol., 253, 102-113. 22

Crowe, D.E., Valley, J.W., Baker, K.L. (1990). Micro-analysis of sulfur-isotope ratios and 23

zonation by laser microprobe. Geochimi. Cosmochim. Acta, 54, 2075-2092. 24

Ding, T., Valkiers, S., Kipphardt, H., De Bievre, P., Taylor, P., Gonfiantini, R., Krouse, R. 25

2001. Calibrated sulfur isotope abundance ratios of three IAEA sulfur isotope reference 26

materials and V-CDT with a reassessment of the atomic weight of sulfur. Geochimi. 27

Cosmochim. Acta, 65, 2433-2437. 28

Durazzo, A., Taylor, L.A. 1982. Exsolution in mss-pentlandite system: textural and genetic 29

implications for Ni-sulfide ores. Mineral. Dep., 17, 313-332. 30

Evans, K.A., Tomkins, A.G., Cliff, J., Fiorentini, M.L. 2014. Insights into subduction zone 31

sulfur recycling from isotopic analysis of eclogite-hosted sulfides. Chem. Geol., 365, 1-19. 32

Farquhar, J., Bao, H., Thiemens, M., 2000. Atmospheric influence of Earth’s earliest sulfur 33

cycle. Sci., 28, 756-758. 34

Farquhar, J., Wing, B.A. 2003. Multiple sulfur isotopes and evolution of the atmosphere. 35

Earth Planet. Sci. Lett., 213, 1-13. 36

32

Farquhar, J., Johnston D. T. and Wing B. A. 2007. Implications of conservation of mass 1

effects on mass-dependent isotope fractionations: influence of network structure on sulfur 2

isotope phase space of dissimilatory sulfate reduction. Geochim. Cosmochim. Acta, 71, 3

5862–5875. 4 5 Farquhar, J., Cliff, J., Zerkle, A.L., Kamyshny, A., Poulton, S.W., Claire, M., Adams, D., 6

Harms, B. 2013. Pathways for Neoarchean pyrite formation constrained by mass-independent 7

sulfur isotopes. Proc. Natl. Acad. Sci., 110, 17638-17643. 8

Fiorentini, M.L., Barnes, S.J., Maier, W.D., Burnham, O.M., Heggie, G.J. 2011. Global 9

variabilitiy in the platinum-group element contents of komatiites. J. Pet., 52, 83-112. 10

Fiorentini, M.L., Barnes, S.J., Lesher, C.M., Heggie, G.J., Keays, R.R., Bernham, O.M. 2010. 11

Platinum group element geochemistry of mineralized and non-mineralized komatiites and 12

basalt. Econ. Geol., 105, 795-823. 13

Fiorentini M.L., Bekker, A., Rouxel, O., Wing, B.A., Maier, W., Rumble, D. 2012a. Multiple 14

sulfur and iron isotope composition of magmatic Ni-Cu-(PGE) sulfide mineralization from 15

Eastern Botswana. Econ. Geol., 107, 105-116. 16

Fiorentini, M.L., Beresford, S., Barley, M., Duuring, P., Bekker, A., Rosengren, N., Cas, R., 17

Hronsky, J. 2012b. District to camp controls on the genesis of komatiite-hosted nickel sulfide 18

deposits, Agnew-Wiluna greenstone belt, Western Australia: Insights from the multiple sulfur 19

isotopes. Econ. Geol., 107, 781-796. 20

Giacometti, F., Evans, K., Rebay, G., Cliff, J., Tomkins, A.G., Rosetti, P., Vaggelli, G., 21

Adams, D.T. 2014. Sulfur isotope evolution in sulfide ores from Western Alps: Assessing the 22

influence of subduction-related metamorphism. Geochem., Geophys., Geosystems, 15, 1-22. 23

Graham, C.M., Valley, J.W. 1992. Sulphur isotope analysis of pyrites. Chem. Geol. – Isotope 24

Geoscience Section, 101, 169-172. 25

Hauri, E.H., Papineau, D., Wang, J., Hillion, F. 2016. High-precision analysis of multiple 26

sulfur isotopes using NanoSIMS. Chem. Geol., 420, 148-161. 27

Heggie, G.J., Fiorentini, M.L., Barnes, S.J., Barley, M.E. 2012a. Maggie Hays Ni deposit: 28

Part 1: Stratigraphic controls on the style of komatitte emplacement in the 2.9 Ga Lake 29

Johnston Greenstone Belt, Yilgarn Craton, Western Australia. Econ. Geol., 107, 797-816. 30

Heggie, G.J., Fiorentini, M.L., Barnes, S.J., Barley, M.E. 2012b. Maggie Hays Ni deposit: 31

Part 2: Nickel mineralization and the spatial distribution of PGE ore-forming signatures in the 32

Maggie Hays Ni System, Lake Johnston Greenstone Belt, Western Australia. Econ. Geol., 33

107, 817-833. 34

Helt, K.M., Williams-Jones, A.E., Clark, J.R., Wing, B.A., Wares, R.P. 2014. Constraints on 35

the genesis of the Archean oxidized, intrusion-related Canadian malartic gold deposit, 36

Quebec, Canada. Econ. Geol., 109, 713-735. 37

Hervig, R.L. 2002. Anomalous fractionation of sulfur isotopes during sputtering. Rapid 38

Communications in Mass Spectrom., 16, 1774-1778. 39

33

Hiebert, R.S., Bekker, A., Wing, B.A., Rouxel, O.J. 2013. The role of paragneiss assimilation 1

in the origin of the Voisey’s Bay Ni-Cu sulfide deposit, Labrador: Multiple S and Fe isotope 2

evidence. Econ. Geol., 108, 1459-1469. 3

Hulston, J.R., Thode, H.G. 1965. Variations in the S33, S34, and S36 contents of meteorites 4

and their relation to chemical and nuclear effect. J. Geophys. Res., 70, 3475-3484. 5

Jamieson, J.W., Wing, B.A., Farquhar, J., Hannington, M.D. 2013. Neoarchean seawater 6

sulphate concentrations from sulphur isotopes in massive sulphide ore. Nat. Geosci., 6, 61-7

64. 8

Johnston, D.T., Farquhar, J., Canfield, D.E., 2007. Sulfur isotope insights into microbial 9

sulfate reduction: when microbes meet models. Geochim. Cosmochim. Acta, 71, 3929–3947. 10

Johnston, D.T., 2011. Multiple sulfur isotopes and the evolution of Earth’s surface sulfur 11

cycle. Earth-Sci. Rev., 106, 161-183. 12

Kelly, D.P., Vaughn, D.J. 1983. Pyrrhotite-pentlandite ore textures: a mechanistic approach. 13

Mineral. Mag., 47, 453-463. 14

Kita, N.T., Huberty, J.M., Kozdon, R., Beard, B.L., Valley, J.W. 2011. High precision SIMS 15

oxygen, sulfur and iron stable isotope analyses of geological materials: accuracy, surface 16

topography and crystal orientation. Surf., Interface, Anal., 43, 427-431. 17

Kozdon, R., Kita, N.T., Huberty, J.M., Fournelle, J.H., Johnson, C.A., Valley, J.W. 2010. In 18

situ sulfur isotope analysis of sulfide minerals by SIMS: Precision and accuracy, with 19

applications to thermometry of 3.5 Ga Pilbara cherts. Chem. Geol., 275, 243-253. 20

Labidi, J., Cartigny, P., Moreira, M. 2013. Non-chondritic sulphur isotope composition of the 21

terrestrial mantle. Nat., 501, 208–211. 22

Leach D.L., Sangster D.F., Kelley K.D., Large, R.R., Garven, G., Allen, C.R., Gutzmer, J., 23

Walters, S.G. 2005. Sediment-hosted lead-zinc deposits: A global perspective, in: Hedenquist 24

J.W., Thompson J.F.H., Goldfarb R.J., and Richards J.P. (Eds.), Econ. Geol. 100th 25

Anniversary Volume, 1905-2005, pp. 561-607. Littleton, CO: Society of Economic 26

Geologists Inc. 27

Lesher, C.M., Groves, D.I. 1986. Controls on the formation of komatiite-associated nickel-28

copper sulfide deposits, in: Friedrich, G.H. (Ed.), Geology and metallogenesis of copper 29

deposits: Proceedings of the Twenty-Seventh International Geological Congress, Berlin, 30

Springer Verlag, pp. 43–62. 31

Lesher, C.M., Burnham, O.M. 2001. Multicomponent elemental and isotopic mixing in Ni-32

Cu-(PGE) ores at Kambalda, Western Australia. Can. Mineral., 39, 421-446. 33

Li, Y., Liu, J. 2006. Calculation of sulfur isotope fractionation in sulfides. Geochim. 34

Cosmochim. Acta 70, 1789-1795. 35

Li, C., Ripley, E.M., Naldrett, A.J. 2009. A new genetic model for the giant Ni-Cu-PGE 36

sulfide deposits associated with the Siberian flood basalts. Econ. Geol., 104, 291-301. 37

34

McCuaig, T.C., Beresford, S., Hronsky, J. 2010. Translating the mineral systems approach 1

into an effective exploration targeting system. Ore Geol. Rev., 38, 128-138. 2

Naldrett, A.C. 1966. The role of sulphurisation in the genesis of iron-nickel sulphide deposits 3

of the Porcupine district, Ontario. Can. Inst. Min. Metall. Trans., 69, 147-155. 4

Naldrett, A.C., Craig, J.R., Kullerud, G. 1967. The central portion of the Fe-Ni-S system and 5

its bearing on pentlandite exsolution in iron-nickel sulfide ores. Econ. Geol., 62, 826-847. 6

O’Neil, J.R. 1986. Theoretical and experimental aspects of isotopic fractionation. In: Valley, 7

J.W., Taylor, H.P., O’Neil, R. (Eds.), Stable Isotopes in High Temperature Geological 8

Processes. Mineralogical Society of America, Washington, pp. 1-40. 9

Ohmoto, H., Rye, R.O. 1979. Isotopes of sulfur and carbon, in: Barnes, H.L. (Ed.), 10

Geochemistry of Hydrothermal Ore Deposits, 2nd edition. J Wiley and Sons, pp. 509-567. 11

Ono, S., Wing, B., Rumble, D., Farquhar, J. 2006a. High precision analysis of all four stable 12

isotopes of sulfur (32S, 33S, 34S and 36S) at nanomole levels using a laser fluorination isotope-13

ratio-monitoring gas chromatography-mass spectrometry. Chem. Geol., 225, 30-39. 14

Ono, S., Wing B., Johnston D., Farquhar J. and Rumble D. 2006b. Mass-dependent 15

fractionation of quadruple stable sulfur isotope system as a new tracer of sulfur 16

biogeochemical cycles. Geochim. Cosmochim. Acta 70, 2238–2252. 17

18

Penniston-Dorland, S.C., Wing, B.A., Nex, P.A.M., Kinnaird, J.A., Farquhar, J., Brown, M., 19

Sharman, E.R. 2008. Multiple sulfur isotopes reveal a magmatic origin for the Platreef 20

platinum group element deposit, Bushveld Complex, South Africa. Geol., 36, 979−982. 21

Peres, P., Kita, N.T., Valley, J.W., Fernandes, F., Schuhmacher, M. 2012. New sample holder 22

geometry for high precision isotope analyses. Surf. Interface Anal., 45, 553-556. 23

Rajamani, V., Prewitt, C.T. 1975. Thermal expansion of the pentlandite structure. Am. Min., 24

60, 39-48. 25

Schuhmacher, M., Fernandes, F., De Chambost, E. 2003. Achieving high reproducibility 26

isotope ratios with the Cameca IMS 1270 in the multicollection mode. Appl. Surf. Sci., 231-27

232, 878-882. 28

Seal, R.R. 2006. Sulfur isotope geochemistry of sulfide minerals. Rev. Mineral. Geochem., 29

61, 633-677. 30

Selvaraja, V., LaFlamme, C., Fiorentini, M. Wing, B., Bui, T.H. Anomalous isotope 31

signatures trace sulfur pathways in magmatic arcs, In review. 32

33

Sharman, E.R. Penniston-Dorland, S.C., Kinnaird, J.A., Nex, P.A.M., Brown, M., Wing, B.A. 34

2013. Primary origin of marginal Ni-Cu-(PGE) mineralization in layered intrusions: Δ33S 35

evidence from The Platreef, Bushveld, South Africa. Econ. Geol., 108, 365-377. 36

Sharman, E.R. Taylor, B.E., Minarik, W.G., Dubé, B., Wing, B.A. 2015. Sulfur isotope and 37

trace element data from ore sulfides in the Noranda district (Abitibi, Canada); implications 38

for volcanogenic massive sulfide deposit genesis. Mineral. Dep., 50, 591-606. 39

35

Shimizu, N., Hart, S.R. 1982. Applications of the ion microprobe to geochemistry and 1

cosmochemistry. Ann. Rev. Earth Planet. Sci., 10, 483-526. 2

Stern, R.A. 2008. An introduction to secondary ion mass spectrometry (SIMS) in geology. 3

Mineralogical Association of Canada Short Course 41, 1-18. 4

Stone, W.E., Beresford, S.W., Archibald, N.J. 2005. Structural setting and shape analysis of 5

nickel sulfide shoots at the Kambalda Dome, Western Australia: implications for deformation 6

and remobilization. Econ. Geol., 100, 1441-1455. 7

Strauss, H. 1997. The isotopic composition of sedimentary sulfur through time. Palaeogeogr., 8

Palaeoclimatology, Palaeoecol., 132, 97-118. 9

Taylor, R., Clark, C., Reddy, S.M., The effect of grain orientation on secondary ion mass 10

spectrometry (SIMS) analysis of rutile. Chem. Geol., 300, 81-87. 11

Thomassot, E., O’Neil, J., Francis, D., Cartigny, P., Wing, B. 2015. Atmospheric record in 12

the Hadean Eon from multiple sulfur isotope measurements in Nuvvuagittuq Greenstone Belt 13

(Nunavik, Quebec). Proc. Nat. Aca. Sci., 112, 707-712. 14

Ushikubo, T., Williford, K.H., Farquhar, J., Johnston, D.T., Van Kranendonk, M.J., Valley, 15

J.W. 2014. Development of in situ four-isotope analysis with multiple Faraday cup detectors 16

by SIMS and application to pyrite grains in a Paleoproterozoic glaciogenic sandstone. Chem. 17

Geol., 383, 86-99. 18

Wacey, D., Kilburn, M.R., Saunders, M., Cliff, J., and Brasier, M.D. 2011. Microfossils of 19

sulfur metabolizing cells in ~3.4 billion year old rocks of Western Australia. Nat. Geosci., 4, 20

698-702. 21

Whitehouse, M. J., Kamber, B.S., Fedo, C.M., Lepland, A. Integrated Pb- and S-isotope 22

investigation of sulphide minerals from the early Archaean of southwest Greenland. Chem. 23

Geol., 222, 112-131. 24

Whitehouse, M., 2013. Multiple sulfur isotope determination by SIMS: Evaluation of 25

reference sulfides for ∆33S with observations and a case study on the determination of ∆36S. 26

Geostand. Geoanal. Res., 37, 19-33. 27

Williford, K.H., Van Kranendonk, M.J., Ushikubo, T., Kozdon, R., Valley, J. 2011. 28

Constraining atmospheric oxygen and seawater sulfate concentrations during 29

Paleoproterozoic glaciation: In situ sulfur three-isotope microanalysis of pyrite from the 30

Turee Creek Group, Western Australia. Geochim. Cosmochim. Acta, 75, 5686-5705. 31

Wing, B. Farquhar, J. 2015. Sulfur isotope homogeneity of lunar mare basalts. Geochim. 32

Cosmochim. Acta, 170, 266-280. 33

Xue, Y., Campbell, I., Ireland, T. Holden, P., Armstrong, R. 2013. No mass-independent 34

sulfur isotope fractionation in auriferous fluids supports a magmatic origin for Archean gold 35

deposits. Geol., 41, 791-794. 36